Embed the Power of Lua into NGINX HTTP servers
11,788
stars
3,973
commits
C
primary language
Sep 7, 2026
updated
ngx_http_lua_module - Embed the power of Lua into Nginx HTTP Servers.
This module is a core component of OpenResty. If you are using this module, then you are essentially using OpenResty :)
This module is not distributed with the Nginx source. See the installation instructions.
Production ready.
This document describes ngx_lua v0.10.29, which was released on Oct 24, 2025.
YouTube video "Hello World HTTP Example with OpenResty/Lua"
YouTube video "Write Your Own Lua Modules in OpenResty/Nginx Applications"
YouTube video "OpenResty's resty Command-Line Utility Demo"
YouTube video "Measure Execution Time of Lua Code Correctly in OpenResty"
YouTube video "Precompile Lua Modules into LuaJIT Bytecode to Speedup OpenResty Startup"
You are welcome to subscribe to our official YouTube channel, OpenResty.
# set search paths for pure Lua external libraries (';;' is the default path):
lua_package_path '/foo/bar/?.lua;/blah/?.lua;;';
# set search paths for Lua external libraries written in C (can also use ';;'):
lua_package_cpath '/bar/baz/?.so;/blah/blah/?.so;;';
server {
location /lua_content {
# MIME type determined by default_type:
default_type 'text/plain';
content_by_lua_block {
ngx.say('Hello,world!')
}
}
location /nginx_var {
# MIME type determined by default_type:
default_type 'text/plain';
# try access /nginx_var?a=hello,world
content_by_lua_block {
ngx.say(ngx.var.arg_a)
}
}
location = /request_body {
client_max_body_size 50k;
client_body_buffer_size 50k;
content_by_lua_block {
ngx.req.read_body() -- explicitly read the req body
local data = ngx.req.get_body_data()
if data then
ngx.say("body data:")
ngx.print(data)
return
end
-- body may get buffered in a temp file:
local file = ngx.req.get_body_file()
if file then
ngx.say("body is in file ", file)
else
ngx.say("no body found")
end
}
}
# transparent non-blocking I/O in Lua via subrequests
# (well, a better way is to use cosockets)
location = /lua {
# MIME type determined by default_type:
default_type 'text/plain';
content_by_lua_block {
local res = ngx.location.capture("/some_other_location")
if res then
ngx.say("status: ", res.status)
ngx.say("body:")
ngx.print(res.body)
end
}
}
location = /foo {
rewrite_by_lua_block {
res = ngx.location.capture("/memc",
{ args = { cmd = "incr", key = ngx.var.uri } }
)
}
proxy_pass http://blah.blah.com;
}
location = /mixed {
rewrite_by_lua_file /path/to/rewrite.lua;
access_by_lua_file /path/to/access.lua;
content_by_lua_file /path/to/content.lua;
}
# use nginx var in code path
# CAUTION: contents in nginx var must be carefully filtered,
# otherwise there'll be great security risk!
location ~ ^/app/([-_a-zA-Z0-9/]+) {
set $path $1;
content_by_lua_file /path/to/lua/app/root/$path.lua;
}
location / {
client_max_body_size 100k;
client_body_buffer_size 100k;
access_by_lua_block {
-- check the client IP address is in our black list
if ngx.var.remote_addr == "132.5.72.3" then
ngx.exit(ngx.HTTP_FORBIDDEN)
end
-- check if the URI contains bad words
if ngx.var.uri and
string.match(ngx.var.request_body, "evil")
then
return ngx.redirect("/terms_of_use.html")
end
-- tests passed
}
# proxy_pass/fastcgi_pass/etc settings
}
}
This module embeds LuaJIT 2.0/2.1 into Nginx. It is a core component of OpenResty. If you are using this module, then you are essentially using OpenResty.
Since version v0.10.16 of this module, the standard Lua
interpreter (also known as "PUC-Rio Lua") is not supported anymore. This
document interchangeably uses the terms "Lua" and "LuaJIT" to refer to the
LuaJIT interpreter.
By leveraging Nginx's subrequests, this module allows the integration of the powerful Lua threads (known as Lua "coroutines") into the Nginx event model.
Unlike Apache's mod_lua and Lighttpd's mod_magnet, Lua code executed using this module can be 100% non-blocking on network traffic as long as the Nginx API for Lua provided by this module is used to handle requests to upstream services such as MySQL, PostgreSQL, Memcached, Redis, or upstream HTTP web services.
At least the following Lua libraries and Nginx modules can be used with this module:
Almost any Nginx modules can be used with this ngx_lua module by means of
ngx.location.capture or
ngx.location.capture_multi but it is
recommended to use those lua-resty-* libraries instead of creating
subrequests to access the Nginx upstream modules because the former is usually
much more flexible and memory-efficient.
The Lua interpreter (also known as "Lua State" or "LuaJIT VM instance") is shared across all the requests in a single Nginx worker process to minimize memory use. Request contexts are segregated using lightweight Lua coroutines.
Loaded Lua modules persist in the Nginx worker process level resulting in a small memory footprint in Lua even when under heavy loads.
This module is plugged into Nginx's "http" subsystem so it can only speak downstream communication protocols in the HTTP family (HTTP 0.9/1.0/1.1/2.0, WebSockets, etc...). If you want to do generic TCP communications with the downstream clients, then you should use the ngx_stream_lua module instead, which offers a compatible Lua API.
Just to name a few:
The possibilities are unlimited as the module allows bringing together various elements within Nginx as well as exposing the power of the Lua language to the user. The module provides the full flexibility of scripting while offering performance levels comparable with native C language programs both in terms of CPU time as well as memory footprint thanks to LuaJIT 2.x.
Other scripting language implementations typically struggle to match this performance level.
The latest version of this module is compatible with the following versions of Nginx:
Nginx cores older than 1.6.0 (exclusive) are not supported.
It is highly recommended to use OpenResty releases which bundle Nginx, ngx_lua (this module), LuaJIT, as well as other powerful companion Nginx modules and Lua libraries.
It is discouraged to build this module with Nginx yourself since it is tricky to set up exactly right.
Note that Nginx, LuaJIT, and OpenSSL official releases have various limitations and long-standing bugs that can cause some of this module's features to be disabled, not work properly, or run slower. Official OpenResty releases are recommended because they bundle OpenResty's optimized LuaJIT 2.1 fork and Nginx/OpenSSL patches.
Alternatively, ngx_lua can be manually compiled into Nginx:
Build the source with this module:
wget 'https://openresty.org/download/nginx-1.19.3.tar.gz'
tar -xzvf nginx-1.19.3.tar.gz
cd nginx-1.19.3/
# tell nginx's build system where to find LuaJIT 2.0:
export LUAJIT_LIB=/path/to/luajit/lib
export LUAJIT_INC=/path/to/luajit/include/luajit-2.0
# tell nginx's build system where to find LuaJIT 2.1:
export LUAJIT_LIB=/path/to/luajit/lib
export LUAJIT_INC=/path/to/luajit/include/luajit-2.1
# Here we assume Nginx is to be installed under /opt/nginx/.
./configure --prefix=/opt/nginx \
--with-ld-opt="-Wl,-rpath,/path/to/luajit/lib" \
--add-module=/path/to/ngx_devel_kit \
--add-module=/path/to/lua-nginx-module
# Note that you may also want to add `./configure` options which are used in your
# current nginx build.
# You can get usually those options using command nginx -V
# you can change the parallelism number 2 below to fit the number of spare CPU cores in your
# machine.
make -j2
make install
# Note that this version of lug-nginx-module not allow to set `lua_load_resty_core off;` any more.
# So, you have to install `lua-resty-core` and `lua-resty-lrucache` manually as below.
cd lua-resty-core
make install PREFIX=/opt/nginx
cd lua-resty-lrucache
make install PREFIX=/opt/nginx
# add necessary `lua_package_path` directive to `nginx.conf`, in the http context
lua_package_path "/opt/nginx/lib/lua/?.lua;;";
Starting from NGINX 1.9.11, you can also compile this module as a dynamic module, by using the --add-dynamic-module=PATH option instead of --add-module=PATH on the
./configure command line above. And then you can explicitly load the module in your nginx.conf via the load_module
directive, for example,
load_module /path/to/modules/ndk_http_module.so; # assuming NDK is built as a dynamic module too
load_module /path/to/modules/ngx_http_lua_module.so;
While building this module either via OpenResty or with the Nginx core, you can define the following C macros via the C compiler options:
NGX_LUA_USE_ASSERT
When defined, will enable assertions in the ngx_lua C code base. Recommended for debugging or testing builds. It can introduce some (small) runtime overhead when enabled. This macro was first introduced in the v0.9.10 release.NGX_LUA_ABORT_AT_PANIC
When the LuaJIT VM panics, ngx_lua will instruct the current nginx worker process to quit gracefully by default. By specifying this C macro, ngx_lua will abort the current nginx worker process (which usually results in a core dump file) immediately. This option is useful for debugging VM panics. This option was first introduced in the v0.9.8 release.To enable one or more of these macros, just pass extra C compiler options to the ./configure script of either Nginx or OpenResty. For instance,
./configure --with-cc-opt="-DNGX_LUA_USE_ASSERT -DNGX_LUA_ABORT_AT_PANIC"
The openresty-en mailing list is for English speakers.
The openresty mailing list is for Chinese speakers.
The code repository of this project is hosted on GitHub at openresty/lua-nginx-module.
Please submit bug reports, wishlists, or patches by
Watch YouTube video "Measure Execution Time of Lua Code Correctly in OpenResty"
As from the v0.5.0rc32 release, all *_by_lua_file configure directives (such as content_by_lua_file) support loading LuaJIT 2.0/2.1 raw bytecode files directly:
/path/to/luajit/bin/luajit -b /path/to/input_file.lua /path/to/output_file.ljbc
The -bg option can be used to include debug information in the LuaJIT bytecode file:
/path/to/luajit/bin/luajit -bg /path/to/input_file.lua /path/to/output_file.ljbc
Please refer to the official LuaJIT documentation on the -b option for more details:
https://luajit.org/running.html#opt_b
Note that the bytecode files generated by LuaJIT 2.1 is not compatible with LuaJIT 2.0, and vice versa. The support for LuaJIT 2.1 bytecode was first added in ngx_lua v0.9.3.
Attempts to load standard Lua 5.1 bytecode files into ngx_lua instances linked to LuaJIT 2.0/2.1 (or vice versa) will result in an Nginx error message such as the one below:
[error] 13909#0: *1 failed to load Lua inlined code: bad byte-code header in /path/to/test_file.luac
Loading bytecode files via the Lua primitives like require and
dofile should always work as expected.
If you want to access the system environment variable, say, foo, in Lua via the standard Lua API os.getenv, then you should also list this environment variable name in your nginx.conf file via the env directive. For example,
env foo;
The HTTP 1.0 protocol does not support chunked output and requires an explicit Content-Length header when the response body is not empty in order to support the HTTP 1.0 keep-alive.
So when a HTTP 1.0 request is made and the lua_http10_buffering directive is turned on, ngx_lua will buffer the
output of ngx.say and ngx.print calls and also postpone sending response headers until all the response body output is received.
At that time ngx_lua can calculate the total length of the body and construct a proper Content-Length header to return to the HTTP 1.0 client.
If the Content-Length response header is set in the running Lua code, however, this buffering will be disabled even if the lua_http10_buffering directive is turned on.
For large streaming output responses, it is important to disable the lua_http10_buffering directive to minimise memory usage.
Note that common HTTP benchmark tools such as ab and http_load issue HTTP 1.0 requests by default.
To force curl to send HTTP 1.0 requests, use the -0 option.
With LuaJIT 2.x, it is possible to statically link the bytecode of pure Lua modules into the Nginx executable.
You can use the luajit executable to compile .lua Lua
module files to .o object files containing the exported bytecode
data, and then link the .o files directly in your Nginx build.
Below is a trivial example to demonstrate this. Consider that we have the following .lua file named foo.lua:
-- foo.lua
local _M = {}
function _M.go()
print("Hello from foo")
end
return _M
And then we compile this .lua file to foo.o file:
/path/to/luajit/bin/luajit -bg foo.lua foo.o
What matters here is the name of the .lua file, which determines how you use this module later on the Lua land. The file name foo.o does not matter at all except the .o file extension (which tells luajit what output format is used). If you want to strip the Lua debug information from the resulting bytecode, you can just specify the -b option above instead of -bg.
Then when building Nginx or OpenResty, pass the --with-ld-opt="foo.o" option to the ./configure script:
./configure --with-ld-opt="/path/to/foo.o" ...
Finally, you can just do the following in any Lua code run by ngx_lua:
local foo = require "foo"
foo.go()
And this piece of code no longer depends on the external foo.lua file any more because it has already been compiled into the nginx executable.
If you want to use dot in the Lua module name when calling require, as in
local foo = require "resty.foo"
then you need to rename the foo.lua file to resty_foo.lua before compiling it down to a .o file with the luajit command-line utility.
It is important to use exactly the same version of LuaJIT when compiling .lua files to .o files as building nginx + ngx_lua. This is because the LuaJIT bytecode format may be incompatible between different LuaJIT versions. When the bytecode format is incompatible, you will see a Lua runtime error saying that the Lua module is not found.
When you have multiple .lua files to compile and link, then just specify their .o files at the same time in the value of the --with-ld-opt option. For instance,
./configure --with-ld-opt="/path/to/foo.o /path/to/bar.o" ...
If you have too many .o files, then it might not be feasible to name them all in a single command. In this case, you can build a static library (or archive) for your .o files, as in
ar rcus libmyluafiles.a *.o
then you can link the myluafiles archive as a whole to your nginx executable:
./configure \
--with-ld-opt="-L/path/to/lib -Wl,--whole-archive -lmyluafiles -Wl,--no-whole-archive"
where /path/to/lib is the path of the directory containing the libmyluafiles.a file. It should be noted that the linker option --whole-archive is required here because otherwise our archive will be skipped because no symbols in our archive are mentioned in the main parts of the nginx executable.
To globally share data among all the requests handled by the same Nginx worker
process, encapsulate the shared data into a Lua module, use the Lua
require builtin to import the module, and then manipulate the
shared data in Lua. This works because required Lua modules are loaded only
once and all coroutines will share the same copy of the module (both its code
and data).
Note that the use of global Lua variables is strongly discouraged, as it may lead to unexpected race conditions between concurrent requests.
Here is a small example on sharing data within an Nginx worker via a Lua module:
-- mydata.lua
local _M = {}
local data = {
dog = 3,
cat = 4,
pig = 5,
}
function _M.get_age(name)
return data[name]
end
return _M
and then accessing it from nginx.conf:
location /lua {
content_by_lua_block {
local mydata = require "mydata"
ngx.say(mydata.get_age("dog"))
}
}
The mydata module in this example will only be loaded and run on the first request to the location /lua,
and all subsequent requests to the same Nginx worker process will use the reloaded instance of the
module as well as the same copy of the data in it, until a HUP signal is sent to the Nginx master process to force a reload.
This data sharing technique is essential for high performance Lua applications based on this module.
Note that this data sharing is on a per-worker basis and not on a per-server basis. That is, when there are multiple Nginx worker processes under an Nginx master, data sharing cannot cross the process boundary between these workers.
It is usually recommended to share read-only data this way. You can also share changeable data among all the concurrent requests of each Nginx worker process as long as there is no nonblocking I/O operations (including ngx.sleep) in the middle of your calculations. As long as you do not give the control back to the Nginx event loop and ngx_lua's light thread scheduler (even implicitly), there can never be any race conditions in between. For this reason, always be very careful when you want to share changeable data on the worker level. Buggy optimizations can easily lead to hard-to-debug race conditions under load.
If server-wide data sharing is required, then use one or more of the following approaches:
memcached, redis, MySQL or PostgreSQL. The OpenResty official releases come with a set of companion Nginx modules and Lua libraries that provide interfaces with these data storage mechanisms.The tcpsock:connect method may indicate success despite connection failures such as with Connection Refused errors.
However, later attempts to manipulate the cosocket object will fail and return the actual error status message generated by the failed connect operation.
This issue is due to limitations in the Nginx event model and only appears to affect Mac OS X.
dofile and require builtins are currently implemented as C functions in LuaJIT 2.0/2.1, if the Lua file being loaded by dofile or require invokes ngx.location.capture*, ngx.exec, ngx.exit, or other API functions requiring yielding in the top-level scope of the Lua file, then the Lua error "attempt to yield across C-call boundary" will be raised. To avoid this, put these calls requiring yielding into your own Lua functions in the Lua file instead of the top-level scope of the file.Care must be taken when importing modules, and this form should be used:
local xxx = require('xxx')
instead of the old deprecated form:
require('xxx')
Here is the reason: by design, the global environment has exactly the same lifetime as the Nginx request handler associated with it. Each request handler has its own set of Lua global variables and that is the idea of request isolation. The Lua module is actually loaded by the first Nginx request handler and is cached by the require() built-in in the package.loaded table for later reference, and the module() builtin used by some Lua modules has the side effect of setting a global variable to the loaded module table. But this global variable will be cleared at the end of the request handler, and every subsequent request handler all has its own (clean) global environment. So one will get Lua exception for accessing the nil value.
The use of Lua global variables is a generally inadvisable in the ngx_lua context as:
It is therefore highly recommended to always declare such within an appropriate local scope instead.
-- Avoid
foo = 123
-- Recommended
local foo = 123
-- Avoid
function foo() return 123 end
-- Recommended
local function foo() return 123 end
To find all instances of Lua global variables in your Lua code, run the lua-releng tool across all .lua source files:
$ lua-releng
Checking use of Lua global variables in file lib/foo/bar.lua ...
1 [1489] SETGLOBAL 7 -1 ; contains
55 [1506] GETGLOBAL 7 -3 ; setvar
3 [1545] GETGLOBAL 3 -4 ; varexpand
The output says that the line 1489 of file lib/foo/bar.lua writes to a global variable named contains, the line 1506 reads from the global variable setvar, and line 1545 reads the global varexpand.
This tool will guarantee that local variables in the Lua module functions are all declared with the local keyword, otherwise a runtime exception will be thrown. It prevents undesirable race conditions while accessing such variables. See Data Sharing within an Nginx Worker for the reasons behind this.
The ngx.location.capture and ngx.location.capture_multi directives cannot capture locations that include the add_before_body, add_after_body, auth_request, echo_location, echo_location_async, echo_subrequest, or echo_subrequest_async directives.
location /foo {
content_by_lua_block {
res = ngx.location.capture("/bar")
}
}
location /bar {
echo_location /blah;
}
location /blah {
echo "Success!";
}
$ curl -i http://example.com/foo
will not work as expected.
Due to internal limitations in the Nginx core, the cosocket API is disabled in the following contexts: set_by_lua*, log_by_lua*, header_filter_by_lua*, and body_filter_by_lua.
The cosockets are currently also disabled in the init_by_lua* and init_worker_by_lua* directive contexts but we may add support for these contexts in the future because there is no limitation in the Nginx core (or the limitation might be worked around).
There exists a workaround, however, when the original context does not need to wait for the cosocket results. That is, creating a zero-delay timer via the ngx.timer.at API and do the cosocket results in the timer handler, which runs asynchronously as to the original context creating the timer.
NOTE Following the v0.9.17 release, this pitfall can be avoided by using the *_by_lua_block {} configuration directives.
PCRE sequences such as \d, \s, or \w, require special attention because in string literals, the backslash character, \, is stripped out by both the Lua language parser and by the Nginx config file parser before processing if not within a *_by_lua_block {} directive. So the following snippet will not work as expected:
# nginx.conf
? location /test {
? content_by_lua '
? local regex = "\d+" -- THIS IS WRONG OUTSIDE OF A *_by_lua_block DIRECTIVE
? local m = ngx.re.match("hello, 1234", regex)
? if m then ngx.say(m[0]) else ngx.say("not matched!") end
? ';
? }
# evaluates to "not matched!"
To avoid this, double escape the backslash:
# nginx.conf
location /test {
content_by_lua '
local regex = "\\\\d+"
local m = ngx.re.match("hello, 1234", regex)
if m then ngx.say(m[0]) else ngx.say("not matched!") end
';
}
# evaluates to "1234"
Here, \\\\d+ is stripped down to \\d+ by the Nginx config file parser and this is further stripped down to \d+ by the Lua language parser before running.
Alternatively, the regex pattern can be presented as a long-bracketed Lua string literal by encasing it in "long brackets", [[...]], in which case backslashes have to only be escaped once for the Nginx config file parser.
# nginx.conf
location /test {
content_by_lua '
local regex = [[\\d+]]
local m = ngx.re.match("hello, 1234", regex)
if m then ngx.say(m[0]) else ngx.say("not matched!") end
';
}
# evaluates to "1234"
Here, [[\\d+]] is stripped down to [[\d+]] by the Nginx config file parser and this is processed correctly.
Note that a longer from of the long bracket, [=[...]=], may be required if the regex pattern contains [...] sequences.
The [=[...]=] form may be used as the default form if desired.
# nginx.conf
location /test {
content_by_lua '
local regex = [=[[0-9]+]=]
local m = ngx.re.match("hello, 1234", regex)
if m then ngx.say(m[0]) else ngx.say("not matched!") end
';
}
# evaluates to "1234"
An alternative approach to escaping PCRE sequences is to ensure that Lua code is placed in external script files and executed using the various *_by_lua_file directives.
With this approach, the backslashes are only stripped by the Lua language parser and therefore only need to be escaped once each.
-- test.lua
local regex = "\\d+"
local m = ngx.re.match("hello, 1234", regex)
if m then ngx.say(m[0]) else ngx.say("not matched!") end
-- evaluates to "1234"
Within external script files, PCRE sequences presented as long-bracketed Lua string literals do not require modification.
-- test.lua
local regex = [[\d+]]
local m = ngx.re.match("hello, 1234", regex)
if m then ngx.say(m[0]) else ngx.say("not matched!") end
-- evaluates to "1234"
As noted earlier, PCRE sequences presented within *_by_lua_block {} directives (available following the v0.9.17 release) do not require modification.
# nginx.conf
location /test {
content_by_lua_block {
local regex = [[\d+]]
local m = ngx.re.match("hello, 1234", regex)
if m then ngx.say(m[0]) else ngx.say("not matched!") end
}
}
# evaluates to "1234"
NOTE You are recommended to use by_lua_file when the Lua code is very long.
Mixing SSI with ngx_lua in the same Nginx request is not supported at all. Just use ngx_lua exclusively. Everything you can do with SSI can be done atop ngx_lua anyway and it can be more efficient when using ngx_lua.
Certain Lua APIs provided by ngx_lua do not work in Nginx's SPDY mode yet: ngx.location.capture, ngx.location.capture_multi, and ngx.req.socket.
Nginx may terminate a request early with (at least):
This means that phases that normally run are skipped, such as the rewrite or access phase. This also means that later phases that are run regardless, e.g. log_by_lua, will not have access to information that is normally set in those phases.
bsdrecv method.ngx_hash_t to optimize the built-in header look-up process for ngx.req.set_header, and etc.ignore_resp_headers, ignore_resp_body, and ignore_resp options to ngx.location.capture and ngx.location.capture_multi methods, to allow micro performance tuning on the user side.stat mode similar to mod_lua.The changes made in every release of this module are listed in the change logs of the OpenResty bundle:
https://openresty.org/#Changes
This module uses .travis.yml as the CI configuration.
You can always check .travis.yml for the latest CI configuration.
For developers, you need to run tests locally. You can use util/run-ci.sh
to easily set up the environment and execute the test suite.
To run the Test from the beginning:
git clone https://github.com/openresty/lua-nginx-module.git
cd lua-nginx-module
bash util/run-ci.sh
The following dependencies are required to run the test suite:
Nginx version >= 1.4.2
Perl modules:
Nginx modules:
The order in which these modules are added during configuration is important because the position of any filter module in the filtering chain determines the final output, for example. The correct adding order is shown above.
3rd-party Lua libraries:
Applications:
See also the developer build script for more details on setting up the testing environment.
To run the whole test suite in the default testing mode:
cd /path/to/lua-nginx-module
export PATH=/path/to/your/nginx/sbin:$PATH
prove -I/path/to/test-nginx/lib -r t
To run specific test files:
cd /path/to/lua-nginx-module
export PATH=/path/to/your/nginx/sbin:$PATH
prove -I/path/to/test-nginx/lib t/002-content.t t/003-errors.t
To run a specific test block in a particular test file, add the line --- ONLY to the test block you want to run, and then use the prove utility to run that .t file.
There are also various testing modes based on mockeagain, valgrind, and etc. Refer to the Test::Nginx documentation for more details for various advanced testing modes. See also the test reports for the Nginx test cluster running on Amazon EC2: https://qa.openresty.org.
This module is licensed under the BSD license.
Copyright (C) 2009-2017, by Xiaozhe Wang (chaoslawful) chaoslawful@gmail.com.
Copyright (C) 2009-2025, by Yichun "agentzh" Zhang (章亦春) agentzh@gmail.com, OpenResty Inc.
All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:
Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
Blog posts:
Other related modules and libraries:
The basic building blocks of scripting Nginx with Lua are directives. Directives are used to specify when the user Lua code is run and how the result will be used. Below is a diagram showing the order in which directives are executed.

syntax: lua_load_resty_core on|off
default: lua_load_resty_core on
context: http
This directive is deprecated since the v0.10.16 release of this
module. The resty.core module from
lua-resty-core is now mandatorily
loaded during the Lua VM initialization. Specifying this directive will have no
effect.
This directive was first introduced in the v0.10.15 release and
used to optionally load the resty.core module.
syntax: lua_capture_error_log size
default: none
context: http
Enables a buffer of the specified size for capturing all the Nginx error log message data (not just those produced
by this module or the Nginx http subsystem, but everything) without touching files or disks.
You can use units like k and m in the size value, as in
lua_capture_error_log 100k;
As a rule of thumb, a 4KB buffer can usually hold about 20 typical error log messages. So do the maths!
This buffer never grows. If it is full, new error log messages will replace the oldest ones in the buffer.
The size of the buffer must be bigger than the maximum length of a single error log message (which is 4K in OpenResty and 2K in stock NGINX).
You can read the messages in the buffer on the Lua land via the get_logs() function of the ngx.errlog module of the lua-resty-core library. This Lua API function will return the captured error log messages and also remove these already read from the global capturing buffer, making room for any new error log data. For this reason, the user should not configure this buffer to be too big if the user read the buffered error log data fast enough.
Note that the log level specified in the standard error_log directive does have effect on this capturing facility. It only captures log messages of a level no lower than the specified log level in the error_log directive. The user can still choose to set an even higher filtering log level on the fly via the Lua API function errlog.set_filter_level. So it is more flexible than the static error_log directive.
It is worth noting that there is no way to capture the debugging logs
without building OpenResty or Nginx with the ./configure
option --with-debug. And enabling debugging logs is
strongly discouraged in production builds due to high overhead.
This directive was first introduced in the v0.10.9 release.
syntax: lua_use_default_type on | off
default: lua_use_default_type on
context: http, server, location, location if
Specifies whether to use the MIME type specified by the default_type directive for the default value of the Content-Type response header. Deactivate this directive if a default Content-Type response header for Lua request handlers is not desired.
This directive is turned on by default.
This directive was first introduced in the v0.9.1 release.
syntax: lua_malloc_trim <request-count>
default: lua_malloc_trim 1000
context: http
Asks the underlying libc runtime library to release its cached free memory back to the operating system every
N requests processed by the Nginx core. By default, N is 1000. You can configure the request count
by using your own numbers. Smaller numbers mean more frequent releases, which may introduce higher CPU time consumption and
smaller memory footprint while larger numbers usually lead to less CPU time overhead and relatively larger memory footprint.
Just tune the number for your own use cases.
Configuring the argument to 0 essentially turns off the periodical memory trimming altogether.
lua_malloc_trim 0; # turn off trimming completely
The current implementation uses an Nginx log phase handler to do the request counting. So the appearance of the
log_subrequest on directives in nginx.conf
may make the counting faster when subrequests are involved. By default, only "main requests" count.
Note that this directive does not affect the memory allocated by LuaJIT's own allocator based on the mmap
system call.
This directive was first introduced in the v0.10.7 release.
syntax: lua_code_cache on | off
default: lua_code_cache on
context: http, server, location, location if
Enables or disables the Lua code cache for Lua code in *_by_lua_file directives (like set_by_lua_file and
content_by_lua_file) and Lua modules.
When turning off, every request served by ngx_lua will run in a separate Lua VM instance, starting from the 0.9.3 release. So the Lua files referenced in set_by_lua_file,
content_by_lua_file, access_by_lua_file,
and etc will not be cached
and all Lua modules used will be loaded from scratch. With this in place, developers can adopt an edit-and-refresh approach.
Please note however, that Lua code written inlined within nginx.conf
such as those specified by set_by_lua, content_by_lua,
access_by_lua, and rewrite_by_lua will not be updated when you edit the inlined Lua code in your nginx.conf file because only the Nginx config file parser can correctly parse the nginx.conf
file and the only way is to reload the config file
by sending a HUP signal or just to restart Nginx.
Even when the code cache is enabled, Lua files which are loaded by dofile or loadfile
in *_by_lua_file cannot be cached (unless you cache the results yourself). Usually you can either use the init_by_lua
or init_by_lua_file directives to load all such files or just make these Lua files true Lua modules
and load them via require.
The ngx_lua module does not support the stat mode available with the
Apache mod_lua module (yet).
Disabling the Lua code cache is strongly discouraged for production use and should only be used during development as it has a significant negative impact on overall performance. For example, the performance of a "hello world" Lua example can drop by an order of magnitude after disabling the Lua code cache.
syntax: lua_thread_cache_max_entries <num>
default: lua_thread_cache_max_entries 1024
context: http
Specifies the maximum number of entries allowed in the worker process level lua thread object cache.
This cache recycles the lua thread GC objects among all our "light threads".
A zero value of <num> disables the cache.
Note that this feature requires OpenResty's LuaJIT with the new C API lua_resetthread.
This feature was first introduced in version v0.10.9.
syntax: lua_regex_cache_max_entries <num>
default: lua_regex_cache_max_entries 1024
context: http
Specifies the maximum number of entries allowed in the worker process level compiled regex cache.
The regular expressions used in ngx.re.match, ngx.re.gmatch, ngx.re.sub, and ngx.re.gsub will be cached within this cache if the regex option o (i.e., compile-once flag) is specified.
The default number of entries allowed is 1024 and when this limit is reached, new regular expressions will not be cached (as if the o option was not specified) and there will be one, and only one, warning in the error.log file:
2011/08/27 23:18:26 [warn] 31997#0: *1 lua exceeding regex cache max entries (1024), ...
If you are using the ngx.re.* implementation of lua-resty-core by loading the resty.core.regex module (or just the resty.core module), then an LRU cache is used for the regex cache being used here.
Do not activate the o option for regular expressions (and/or replace string arguments for ngx.re.sub and ngx.re.gsub) that are generated on the fly and give rise to infinite variations to avoid hitting the specified limit.
syntax: lua_regex_match_limit <num>
default: lua_regex_match_limit 0
context: http
Specifies the "match limit" used by the PCRE library when executing the ngx.re API. To quote the PCRE manpage, "the limit ... has the effect of limiting the amount of backtracking that can take place."
When the limit is hit, the error string "pcre_exec() failed: -8" will be returned by the ngx.re API functions on the Lua land.
When setting the limit to 0, the default "match limit" when compiling the PCRE library is used. And this is the default value of this directive.
This directive was first introduced in the v0.8.5 release.
syntax: lua_package_path <lua-style-path-str>
default: The content of LUA_PATH environment variable or Lua's compiled-in defaults.
context: http
Sets the Lua module search path used by scripts specified by set_by_lua,
content_by_lua and others. The path string is in standard Lua path form, and ;;
can be used to stand for the original search paths.
As from the v0.5.0rc29 release, the special notation $prefix or ${prefix} can be used in the search path string to indicate the path of the server prefix usually determined by the -p PATH command-line option while starting the Nginx server.
syntax: lua_package_cpath <lua-style-cpath-str>
default: The content of LUA_CPATH environment variable or Lua's compiled-in defaults.
context: http
Sets the Lua C-module search path used by scripts specified by set_by_lua,
content_by_lua and others. The cpath string is in standard Lua cpath form, and ;;
can be used to stand for the original cpath.
As from the v0.5.0rc29 release, the special notation $prefix or ${prefix} can be used in the search path string to indicate the path of the server prefix usually determined by the -p PATH command-line option while starting the Nginx server.
syntax: init_by_lua <lua-script-str>
context: http
phase: loading-config
NOTE Use of this directive is discouraged following the v0.9.17 release. Use the init_by_lua_block directive instead.
Similar to the init_by_lua_block directive, but accepts the Lua source directly in an Nginx string literal (which requires special character escaping).
For instance,
init_by_lua '
print("I need no extra escaping here, for example: \r\nblah")
'
This directive was first introduced in the v0.5.5 release.
syntax: init_by_lua_block { lua-script }
context: http
phase: loading-config
When Nginx receives the HUP signal and starts reloading the config file, the Lua VM will also be re-created and init_by_lua_block will run again on the new Lua VM. In case that the lua_code_cache directive is turned off (default on), the init_by_lua_block handler will run upon every request because in this special mode a standalone Lua VM is always created for each request.
Usually you can pre-load Lua modules at server start-up by means of this hook and take advantage of modern operating systems' copy-on-write (COW) optimization. Here is an example for pre-loading Lua modules:
# this runs before forking out nginx worker processes:
init_by_lua_block { require "cjson" }
server {
location = /api {
content_by_lua_block {
-- the following require() will just return
-- the already loaded module from package.loaded:
ngx.say(require "cjson".encode{dog = 5, cat = 6})
}
}
}
You can also initialize the lua_shared_dict shm storage at this phase. Here is an example for this:
lua_shared_dict dogs 1m;
init_by_lua_block {
local dogs = ngx.shared.dogs
dogs:set("Tom", 56)
}
server {
location = /api {
content_by_lua_block {
local dogs = ngx.shared.dogs
ngx.say(dogs:get("Tom"))
}
}
}
But note that, the lua_shared_dict's shm storage will not be cleared through a config reload (via the HUP signal, for example). So if you do not want to re-initialize the shm storage in your init_by_lua_block code in this case, then you just need to set a custom flag in the shm storage and always check the flag in your init_by_lua_block code.
Because the Lua code in this context runs before Nginx forks its worker processes (if any), data or code loaded here will enjoy the Copy-on-write (COW) feature provided by many operating systems among all the worker processes, thus saving a lot of memory.
Do not initialize your own Lua global variables in this context because use of Lua global variables have performance penalties and can lead to global namespace pollution (see the Lua Variable Scope section for more details). The recommended way is to use proper Lua module files (but do not use the standard Lua function module() to define Lua modules because it pollutes the global namespace as well) and call require() to load your own module files in init_by_lua_block or other contexts (require() does cache the loaded Lua modules in the global package.loaded table in the Lua registry so your modules will only loaded once for the whole Lua VM instance).
Only a small set of the Nginx API for Lua is supported in this context:
More Nginx APIs for Lua may be supported in this context upon future user requests.
Basically you can safely use Lua libraries that do blocking I/O in this very context because blocking the master process during server start-up is completely okay. Even the Nginx core does blocking I/O (at least on resolving upstream's host names) at the configure-loading phase.
You should be very careful about potential security vulnerabilities in your Lua code registered in this context because the Nginx master process is often run under the root account.
This directive was first introduced in the v0.9.17 release.
See also the following blog posts for more details on OpenResty and Nginx's shared memory zones:
syntax: init_by_lua_file <path-to-lua-script-file>
context: http
phase: loading-config
Equivalent to init_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code or LuaJIT bytecode to be executed.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
This directive was first introduced in the v0.5.5 release.
syntax: init_worker_by_lua <lua-script-str>
context: http
phase: starting-worker
NOTE Use of this directive is discouraged following the v0.9.17 release. Use the init_worker_by_lua_block directive instead.
Similar to the init_worker_by_lua_block directive, but accepts the Lua source directly in an Nginx string literal (which requires special character escaping).
For instance,
init_worker_by_lua '
print("I need no extra escaping here, for example: \r\nblah")
';
This directive was first introduced in the v0.9.5 release.
This hook no longer runs in the cache manager and cache loader processes since the v0.10.12 release.
syntax: init_worker_by_lua_block { lua-script }
context: http
phase: starting-worker
Runs the specified Lua code upon every Nginx worker process's startup when the master process is enabled. When the master process is disabled, this hook will just run after init_by_lua*.
This hook is often used to create per-worker reoccurring timers (via the ngx.timer.at Lua API), either for backend health-check or other timed routine work. Below is an example,
init_worker_by_lua_block {
local delay = 3 -- in seconds
local new_timer = ngx.timer.at
local log = ngx.log
local ERR = ngx.ERR
local check
check = function(premature)
if not premature then
-- do the health check or other routine work
local ok, err = new_timer(delay, check)
if not ok then
log(ERR, "failed to create timer: ", err)
return
end
end
-- do something in timer
end
local hdl, err = new_timer(delay, check)
if not hdl then
log(ERR, "failed to create timer: ", err)
return
end
-- other job in init_worker_by_lua
}
This directive was first introduced in the v0.9.17 release.
This hook no longer runs in the cache manager and cache loader processes since the v0.10.12 release.
syntax: init_worker_by_lua_file <lua-file-path>
context: http
phase: starting-worker
Similar to init_worker_by_lua_block, but accepts the file path to a Lua source file or Lua bytecode file.
This directive was first introduced in the v0.9.5 release.
This hook no longer runs in the cache manager and cache loader processes since the v0.10.12 release.
syntax: exit_worker_by_lua_block { lua-script }
context: http
phase: exiting-worker
Runs the specified Lua code upon every Nginx worker process's exit when the master process is enabled. When the master process is disabled, this hook will run before the Nginx process exits.
This hook is often used to release resources allocated by each worker (e.g. resources allocated by init_worker_by_lua*), or to prevent workers from exiting abnormally.
For example,
exit_worker_by_lua_block {
print("log from exit_worker_by_lua_block")
}
It's not allowed to create a timer (even a 0-delay timer) here since it runs after all timers have been processed.
This directive was first introduced in the v0.10.18 release.
syntax: exit_worker_by_lua_file <path-to-lua-script-file>
context: http
phase: exiting-worker
Similar to exit_worker_by_lua_block, but accepts the file path to a Lua source file or Lua bytecode file.
This directive was first introduced in the v0.10.18 release.
syntax: set_by_lua $res <lua-script-str> [$arg1 $arg2 ...]
context: server, server if, location, location if
phase: rewrite
NOTE Use of this directive is discouraged following the v0.9.17 release. Use the set_by_lua_block directive instead.
Similar to the set_by_lua_block directive, but accepts the Lua source directly in an Nginx string literal (which requires special character escaping), and
For example,
set_by_lua $res ' return 32 + math.cos(32) ';
# $res now has the value "32.834223360507" or alike.
As from the v0.5.0rc29 release, Nginx variable interpolation is disabled in the <lua-script-str> argument of this directive and therefore, the dollar sign character ($) can be used directly.
This directive requires the ngx_devel_kit module.
syntax: set_by_lua_block $res { lua-script }
context: server, server if, location, location if
phase: rewrite
Executes code specified inside a pair of curly braces ({}), and returns string output to $res.
The code inside a pair of curly braces ({}) can make API calls and can retrieve input arguments from the ngx.arg table (index starts from 1 and increases sequentially).
This directive is designed to execute short, fast running code blocks as the Nginx event loop is blocked during code execution. Time consuming code sequences should therefore be avoided.
This directive is implemented by injecting custom commands into the standard ngx_http_rewrite_module's command list. Because ngx_http_rewrite_module does not support nonblocking I/O in its commands, Lua APIs requiring yielding the current Lua "light thread" cannot work in this directive.
At least the following API functions are currently disabled within the context of set_by_lua_block:
In addition, note that this directive can only write out a value to a single Nginx variable at a time. However, a workaround is possible using the ngx.var.VARIABLE interface.
location /foo {
set $diff ''; # we have to predefine the $diff variable here
set_by_lua_block $sum {
local a = 32
local b = 56
ngx.var.diff = a - b -- write to $diff directly
return a + b -- return the $sum value normally
}
echo "sum = $sum, diff = $diff";
}
This directive can be freely mixed with all directives of the ngx_http_rewrite_module, set-misc-nginx-module, and array-var-nginx-module modules. All of these directives will run in the same order as they appear in the config file.
set $foo 32;
set_by_lua_block $bar { return tonumber(ngx.var.foo) + 1 }
set $baz "bar: $bar"; # $baz == "bar: 33"
No special escaping is required in the Lua code block.
This directive requires the ngx_devel_kit module.
This directive was first introduced in the v0.9.17 release.
syntax: set_by_lua_file $res <path-to-lua-script-file> [$arg1 $arg2 ...]
context: server, server if, location, location if
phase: rewrite
Equivalent to set_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or, as from the v0.5.0rc32 release, the LuaJIT bytecode to be executed.
Nginx variable interpolation is supported in the <path-to-lua-script-file> argument string of this directive. But special care must be taken for injection attacks.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
When the Lua code cache is turned on (by default), the user code is loaded once at the first request and cached
and the Nginx config must be reloaded each time the Lua source file is modified.
The Lua code cache can be temporarily disabled during development by
switching lua_code_cache off in nginx.conf to avoid reloading Nginx.
This directive requires the ngx_devel_kit module.
syntax: precontent_by_lua_block { lua-script }
context: http, server, location, location if
phase: precontent tail
Acts as a precontent phase handler and executes Lua code string specified in { <lua-script } for every request.
The Lua code may make API calls and is executed as a new spawned coroutine in an independent global environment (i.e. a sandbox).
Note that this handler always runs after the standard ngx_http_mirror_module and ngx_http_try_files_module. For example:
location /images/ {
try_files $uri /images/default.gif;
precontent_by_lua_block {
ngx.log(ngx.NOTICE, "file found")
}
}
location = /images/default.gif {
expires 30s;
precontent_by_lua_block {
ngx.log(ngx.NOTICE, "file not found, use default.gif instead")
}
}
That is, if a request for /images/foo.jpg comes in and the file does not exist, the request will be internally redirected to /images/default.gif before precontent_by_lua_block, and then the precontent_by_lua_block in new location will run and log "file not found, use default.gif instead".
You can use precontent_by_lua_block to perform some preparatory functions after the access phase handler but before the proxy or other content handler. Especially some functions that cannot be performed in balancer_by_lua_block.
you can use the precontent_by_lua_no_postpone directive to control when to run this handler inside the "precontent" request-processing phase of Nginx.
syntax: precontent_by_lua_file <path-to-lua-script-file>
context: http, server, location, location if
phase: precontent tail
Equivalent to precontent_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or, as from the v0.5.0rc32 release, the LuaJIT bytecode to be executed.
Nginx variables can be used in the <path-to-lua-script-file> string to provide flexibility. This however carries some risks and is not ordinarily recommended.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
When the Lua code cache is turned on (by default), the user code is loaded once at the first request and cached
and the Nginx config must be reloaded each time the Lua source file is modified.
The Lua code cache can be temporarily disabled during development by switching lua_code_cache off in nginx.conf to avoid repeatedly reloading Nginx.
Nginx variables are supported in the file path for dynamic dispatch just as in content_by_lua_file.
But be very careful about malicious user inputs and always carefully validate or filter out the user-supplied path components.
syntax: content_by_lua <lua-script-str>
context: location, location if
phase: content
NOTE Use of this directive is discouraged following the v0.9.17 release. Use the content_by_lua_block directive instead.
Similar to the content_by_lua_block directive, but accepts the Lua source directly in an Nginx string literal (which requires special character escaping).
For instance,
content_by_lua '
ngx.say("I need no extra escaping here, for example: \r\nblah")
';
syntax: content_by_lua_block { lua-script }
context: location, location if
phase: content
For instance,
content_by_lua_block {
ngx.say("I need no extra escaping here, for example: \r\nblah")
}
Acts as a "content handler" and executes Lua code string specified in { lua-script } for every request.
The Lua code may make API calls and is executed as a new spawned coroutine in an independent global environment (i.e. a sandbox).
Do not use this directive and other content handler directives in the same location. For example, this directive and the proxy_pass directive should not be used in the same location.
This directive was first introduced in the v0.9.17 release.
syntax: content_by_lua_file <path-to-lua-script-file>
context: location, location if
phase: content
Equivalent to content_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or, as from the v0.5.0rc32 release, the LuaJIT bytecode to be executed.
If the file is not found, a 404 Not Found status code will be returned, and a 503 Service Temporarily Unavailable status code will be returned in case of errors in reading other files.
Nginx variables can be used in the <path-to-lua-script-file> string to provide flexibility. This however carries some risks and is not ordinarily recommended.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
When the Lua code cache is turned on (by default), the user code is loaded once at the first request and cached
and the Nginx config must be reloaded each time the Lua source file is modified.
The Lua code cache can be temporarily disabled during development by
switching lua_code_cache off in nginx.conf to avoid reloading Nginx.
Nginx variables are supported in the file path for dynamic dispatch, for example:
# CAUTION: contents in nginx var must be carefully filtered,
# otherwise there'll be great security risk!
location ~ ^/app/([-_a-zA-Z0-9/]+) {
set $path $1;
content_by_lua_file /path/to/lua/app/root/$path.lua;
}
But be very careful about malicious user inputs and always carefully validate or filter out the user-supplied path components.
syntax: server_rewrite_by_lua_block { lua-script }
context: http, server
phase: server rewrite
Acts as a server rewrite phase handler and executes Lua code string specified in { lua-script } for every request.
The Lua code may make API calls and is executed as a new spawned coroutine in an independent global environment (i.e. a sandbox).
server {
...
server_rewrite_by_lua_block {
ngx.ctx.a = "server_rewrite_by_lua_block in http"
}
location /lua {
content_by_lua_block {
ngx.say(ngx.ctx.a)
ngx.log(ngx.INFO, ngx.ctx.a)
}
}
}
Just as any other rewrite phase handlers, server_rewrite_by_lua_block also runs in subrequests.
server {
server_rewrite_by_lua_block {
ngx.log(ngx.INFO, "is_subrequest:", ngx.is_subrequest)
}
location /lua {
content_by_lua_block {
local res = ngx.location.capture("/sub")
ngx.print(res.body)
}
}
location /sub {
content_by_lua_block {
ngx.say("OK")
}
}
}
Note that when calling ngx.exit(ngx.OK) within a server_rewrite_by_lua_block handler, the Nginx request processing control flow will still continue to the content handler. To terminate the current request from within a server_rewrite_by_lua_block handler, call ngx.exit with status >= 200 (ngx.HTTP_OK) and status < 300 (ngx.HTTP_SPECIAL_RESPONSE) for successful quits and ngx.exit(ngx.HTTP_INTERNAL_SERVER_ERROR) (or its friends) for failures.
server_rewrite_by_lua_block {
ngx.exit(503)
}
location /bar {
...
# never exec
}
syntax: server_rewrite_by_lua_file <path-to-lua-script-file>
context: http, server
phase: server rewrite
Equivalent to server_rewrite_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or, as from the v0.10.22 release, the LuaJIT bytecode to be executed.
Nginx variables can be used in the <path-to-lua-script-file> string to provide flexibility. This however carries some risks and is not ordinarily recommended.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
When the Lua code cache is turned on (by default), the user code is loaded once at the first request and cached and the Nginx config must be reloaded each time the Lua source file is modified. The Lua code cache can be temporarily disabled during development by switching lua_code_cache off in nginx.conf to avoid reloading Nginx.
syntax: rewrite_by_lua <lua-script-str>
context: http, server, location, location if
phase: rewrite tail
NOTE Use of this directive is discouraged following the v0.9.17 release. Use the rewrite_by_lua_block directive instead.
Similar to the rewrite_by_lua_block directive, but accepts the Lua source directly in an Nginx string literal (which requires special character escaping).
For instance,
rewrite_by_lua '
do_something("hello, world!\nhiya\n")
';
syntax: rewrite_by_lua_block { lua-script }
context: http, server, location, location if
phase: rewrite tail
Acts as a rewrite phase handler and executes Lua code string specified in { lua-script } for every request.
The Lua code may make API calls and is executed as a new spawned coroutine in an independent global environment (i.e. a sandbox).
Note that this handler always runs after the standard ngx_http_rewrite_module. So the following will work as expected:
location /foo {
set $a 12; # create and initialize $a
set $b ""; # create and initialize $b
rewrite_by_lua_block {
ngx.var.b = tonumber(ngx.var.a) + 1
}
echo "res = $b";
}
because set $a 12 and set $b "" run before rewrite_by_lua_block.
On the other hand, the following will not work as expected:
? location /foo {
? set $a 12; # create and initialize $a
? set $b ''; # create and initialize $b
? rewrite_by_lua_block {
? ngx.var.b = tonumber(ngx.var.a) + 1
? }
? if ($b = '13') {
? rewrite ^ /bar redirect;
? break;
? }
?
? echo "res = $b";
? }
because if runs before rewrite_by_lua_block even if it is placed after rewrite_by_lua_block in the config.
The right way of doing this is as follows:
location /foo {
set $a 12; # create and initialize $a
set $b ''; # create and initialize $b
rewrite_by_lua_block {
ngx.var.b = tonumber(ngx.var.a) + 1
if tonumber(ngx.var.b) == 13 then
return ngx.redirect("/bar")
end
}
echo "res = $b";
}
Note that the ngx_eval module can be approximated by using rewrite_by_lua_block. For example,
location / {
eval $res {
proxy_pass http://foo.com/check-spam;
}
if ($res = 'spam') {
rewrite ^ /terms-of-use.html redirect;
}
fastcgi_pass ...;
}
can be implemented in ngx_lua as:
location = /check-spam {
internal;
proxy_pass http://foo.com/check-spam;
}
location / {
rewrite_by_lua_block {
local res = ngx.location.capture("/check-spam")
if res.body == "spam" then
return ngx.redirect("/terms-of-use.html")
end
}
fastcgi_pass ...;
}
Just as any other rewrite phase handlers, rewrite_by_lua_block also runs in subrequests.
Note that when calling ngx.exit(ngx.OK) within a rewrite_by_lua_block handler, the Nginx request processing control flow will still continue to the content handler. To terminate the current request from within a rewrite_by_lua_block handler, call ngx.exit with status >= 200 (ngx.HTTP_OK) and status < 300 (ngx.HTTP_SPECIAL_RESPONSE) for successful quits and ngx.exit(ngx.HTTP_INTERNAL_SERVER_ERROR) (or its friends) for failures.
If the ngx_http_rewrite_module's rewrite directive is used to change the URI and initiate location re-lookups (internal redirections), then any rewrite_by_lua_block or rewrite_by_lua_file_block code sequences within the current location will not be executed. For example,
location /foo {
rewrite ^ /bar;
rewrite_by_lua_block {
ngx.exit(503)
}
}
location /bar {
...
}
Here the Lua code ngx.exit(503) will never run. This will be the case if rewrite ^ /bar last is used as this will similarly initiate an internal redirection. If the break modifier is used instead, there will be no internal redirection and the rewrite_by_lua_block code will be executed.
The rewrite_by_lua_block code will always run at the end of the rewrite request-processing phase unless rewrite_by_lua_no_postpone is turned on.
This directive was first introduced in the v0.9.17 release.
syntax: rewrite_by_lua_file <path-to-lua-script-file>
context: http, server, location, location if
phase: rewrite tail
Equivalent to rewrite_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or, as from the v0.5.0rc32 release, the LuaJIT bytecode to be executed.
Nginx variables can be used in the <path-to-lua-script-file> string to provide flexibility. This however carries some risks and is not ordinarily recommended.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
When the Lua code cache is turned on (by default), the user code is loaded once at the first request and cached and the Nginx config must be reloaded each time the Lua source file is modified. The Lua code cache can be temporarily disabled during development by switching lua_code_cache off in nginx.conf to avoid reloading Nginx.
The rewrite_by_lua_file code will always run at the end of the rewrite request-processing phase unless rewrite_by_lua_no_postpone is turned on.
Nginx variables are supported in the file path for dynamic dispatch just as in content_by_lua_file.
syntax: access_by_lua <lua-script-str>
context: http, server, location, location if
phase: access tail
NOTE Use of this directive is discouraged following the v0.9.17 release. Use the access_by_lua_block directive instead.
Similar to the access_by_lua_block directive, but accepts the Lua source directly in an Nginx string literal (which requires special character escaping).
For instance,
access_by_lua '
do_something("hello, world!\nhiya\n")
';
syntax: access_by_lua_block { lua-script }
context: http, server, location, location if
phase: access tail
Acts as an access phase handler and executes Lua code string specified in { <lua-script } for every request.
The Lua code may make API calls and is executed as a new spawned coroutine in an independent global environment (i.e. a sandbox).
Note that this handler always runs after the standard ngx_http_access_module. So the following will work as expected:
location / {
deny 192.168.1.1;
allow 192.168.1.0/24;
allow 10.1.1.0/16;
deny all;
access_by_lua_block {
local res = ngx.location.capture("/mysql", { ... })
...
}
# proxy_pass/fastcgi_pass/...
}
That is, if a client IP address is in the blacklist, it will be denied before the MySQL query for more complex authentication is executed by access_by_lua_block.
Note that the ngx_auth_request module can be approximated by using access_by_lua_block:
location / {
auth_request /auth;
# proxy_pass/fastcgi_pass/postgres_pass/...
}
can be implemented in ngx_lua as:
location / {
access_by_lua_block {
local res = ngx.location.capture("/auth")
if res.status == ngx.HTTP_OK then
return
end
if res.status == ngx.HTTP_FORBIDDEN then
ngx.exit(res.status)
end
ngx.exit(ngx.HTTP_INTERNAL_SERVER_ERROR)
}
# proxy_pass/fastcgi_pass/postgres_pass/...
}
As with other access phase handlers, access_by_lua_block will not run in subrequests.
Note that when calling ngx.exit(ngx.OK) within a access_by_lua_block handler, the Nginx request processing control flow will still continue to the content handler. To terminate the current request from within a access_by_lua_block handler, call ngx.exit with status >= 200 (ngx.HTTP_OK) and status < 300 (ngx.HTTP_SPECIAL_RESPONSE) for successful quits and ngx.exit(ngx.HTTP_INTERNAL_SERVER_ERROR) (or its friends) for failures.
Starting from the v0.9.20 release, you can use the access_by_lua_no_postpone
directive to control when to run this handler inside the "access" request-processing phase
of Nginx.
This directive was first introduced in the v0.9.17 release.
syntax: access_by_lua_file <path-to-lua-script-file>
context: http, server, location, location if
phase: access tail
Equivalent to access_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or, as from the v0.5.0rc32 release, the LuaJIT bytecode to be executed.
Nginx variables can be used in the <path-to-lua-script-file> string to provide flexibility. This however carries some risks and is not ordinarily recommended.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
When the Lua code cache is turned on (by default), the user code is loaded once at the first request and cached
and the Nginx config must be reloaded each time the Lua source file is modified.
The Lua code cache can be temporarily disabled during development by switching lua_code_cache off in nginx.conf to avoid repeatedly reloading Nginx.
Nginx variables are supported in the file path for dynamic dispatch just as in content_by_lua_file.
syntax: header_filter_by_lua <lua-script-str>
context: http, server, location, location if
phase: output-header-filter
NOTE Use of this directive is discouraged following the v0.9.17 release. Use the header_filter_by_lua_block directive instead.
Similar to the header_filter_by_lua_block directive, but accepts the Lua source directly in an Nginx string literal (which requires special character escaping).
For instance,
header_filter_by_lua '
ngx.header["content-length"] = nil
';
This directive was first introduced in the v0.2.1rc20 release.
syntax: header_filter_by_lua_block { lua-script }
context: http, server, location, location if
phase: output-header-filter
Uses Lua code specified in { lua-script } to define an output header filter.
Note that the following API functions are currently disabled within this context:
Here is an example of overriding a response header (or adding one if absent) in our Lua header filter:
location / {
proxy_pass http://mybackend;
header_filter_by_lua_block {
ngx.header.Foo = "blah"
}
}
This directive was first introduced in the v0.9.17 release.
syntax: header_filter_by_lua_file <path-to-lua-script-file>
context: http, server, location, location if
phase: output-header-filter
Equivalent to header_filter_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or as from the v0.5.0rc32 release, the LuaJIT bytecode to be executed.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
This directive was first introduced in the v0.2.1rc20 release.
syntax: body_filter_by_lua <lua-script-str>
context: http, server, location, location if
phase: output-body-filter
NOTE Use of this directive is discouraged following the v0.9.17 release. Use the body_filter_by_lua_block directive instead.
Similar to the body_filter_by_lua_block directive, but accepts the Lua source directly in an Nginx string literal (which requires special character escaping).
For instance,
body_filter_by_lua '
local data, eof = ngx.arg[1], ngx.arg[2]
';
This directive was first introduced in the v0.5.0rc32 release.
syntax: body_filter_by_lua_block { lua-script-str }
context: http, server, location, location if
phase: output-body-filter
Uses Lua code specified in { lua-script } to define an output body filter.
The input data chunk is passed via ngx.arg[1] (as a Lua string value) and the "eof" flag indicating the end of the response body data stream is passed via ngx.arg[2] (as a Lua boolean value).
Behind the scene, the "eof" flag is just the last_buf (for main requests) or last_in_chain (for subrequests) flag of the Nginx chain link buffers. (Before the v0.7.14 release, the "eof" flag does not work at all in subrequests.)
The output data stream can be aborted immediately by running the following Lua statement:
return ngx.ERROR
This will truncate the response body and usually result in incomplete and also invalid responses.
The Lua code can pass its own modified version of the input data chunk to the downstream Nginx output body filters by overriding ngx.arg[1] with a Lua string or a Lua table of strings. For example, to transform all the lowercase letters in the response body, we can just write:
location / {
proxy_pass http://mybackend;
body_filter_by_lua_block {
ngx.arg[1] = string.upper(ngx.arg[1])
}
}
When setting nil or an empty Lua string value to ngx.arg[1], no data chunk will be passed to the downstream Nginx output filters at all.
Likewise, new "eof" flag can also be specified by setting a boolean value to ngx.arg[2]. For example,
location /t {
echo hello world;
echo hiya globe;
body_filter_by_lua_block {
local chunk = ngx.arg[1]
if string.match(chunk, "hello") then
ngx.arg[2] = true -- new eof
return
end
-- just throw away any remaining chunk data
ngx.arg[1] = nil
}
}
Then GET /t will just return the output
hello world
That is, when the body filter sees a chunk containing the word "hello", then it will set the "eof" flag to true immediately, resulting in truncated but still valid responses.
When the Lua code may change the length of the response body, then it is required to always clear out the Content-Length response header (if any) in a header filter to enforce streaming output, as in
location /foo {
# fastcgi_pass/proxy_pass/...
header_filter_by_lua_block {
ngx.header.content_length = nil
}
body_filter_by_lua_block {
ngx.arg[1] = string.len(ngx.arg[1]) .. "\n"
}
}
Note that the following API functions are currently disabled within this context due to the limitations in Nginx output filter's current implementation:
Nginx output filters may be called multiple times for a single request because response body may be delivered in chunks. Thus, the Lua code specified by in this directive may also run multiple times in the lifetime of a single HTTP request.
This directive was first introduced in the v0.9.17 release.
syntax: body_filter_by_lua_file <path-to-lua-script-file>
context: http, server, location, location if
phase: output-body-filter
Equivalent to body_filter_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or, as from the v0.5.0rc32 release, the LuaJIT bytecode to be executed.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
This directive was first introduced in the v0.5.0rc32 release.
syntax: log_by_lua <lua-script-str>
context: http, server, location, location if
phase: log
NOTE Use of this directive is discouraged following the v0.9.17 release. Use the log_by_lua_block directive instead.
Similar to the log_by_lua_block directive, but accepts the Lua source directly in an Nginx string literal (which requires special character escaping).
For instance,
log_by_lua '
print("I need no extra escaping here, for example: \r\nblah")
';
This directive was first introduced in the v0.5.0rc31 release.
syntax: log_by_lua_block { lua-script }
context: http, server, location, location if
phase: log
Runs the Lua source code inlined as the { lua-script } at the log request processing phase. This does not replace the current access logs, but runs before.
Note that the following API functions are currently disabled within this context:
Here is an example of gathering average data for $upstream_response_time:
lua_shared_dict log_dict 5M;
server {
location / {
proxy_pass http://mybackend;
log_by_lua_block {
local log_dict = ngx.shared.log_dict
local upstream_time = tonumber(ngx.var.upstream_response_time)
local sum = log_dict:get("upstream_time-sum") or 0
sum = sum + upstream_time
log_dict:set("upstream_time-sum", sum)
local newval, err = log_dict:incr("upstream_time-nb", 1)
if not newval and err == "not found" then
log_dict:add("upstream_time-nb", 0)
log_dict:incr("upstream_time-nb", 1)
end
}
}
location = /status {
content_by_lua_block {
local log_dict = ngx.shared.log_dict
local sum = log_dict:get("upstream_time-sum")
local nb = log_dict:get("upstream_time-nb")
if nb and sum then
ngx.say("average upstream response time: ", sum / nb,
" (", nb, " reqs)")
else
ngx.say("no data yet")
end
}
}
}
This directive was first introduced in the v0.9.17 release.
syntax: log_by_lua_file <path-to-lua-script-file>
context: http, server, location, location if
phase: log
Equivalent to log_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or, as from the v0.5.0rc32 release, the LuaJIT bytecode to be executed.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
This directive was first introduced in the v0.5.0rc31 release.
syntax: balancer_by_lua_block { lua-script }
context: upstream
phase: content
This directive runs Lua code as an upstream balancer for any upstream entities defined
by the upstream {} configuration block.
For instance,
upstream foo {
server 127.0.0.1;
balancer_by_lua_block {
-- use Lua to do something interesting here
-- as a dynamic balancer
}
}
server {
location / {
proxy_pass http://foo;
}
}
The resulting Lua load balancer can work with any existing Nginx upstream modules like ngx_proxy and ngx_fastcgi.
Also, the Lua load balancer can work with the standard upstream connection pool mechanism,
i.e., the standard keepalive directive.
Just ensure that the keepalive directive
is used after this balancer_by_lua_block directive in a single upstream {} configuration block.
The Lua load balancer can totally ignore the list of servers defined in the upstream {} block
and select peer from a completely dynamic server list (even changing per request) via the
ngx.balancer module
from the lua-resty-core library.
The Lua code handler registered by this directive might get called more than once in a single downstream request when the Nginx upstream mechanism retries the request on conditions specified by directives like the proxy_next_upstream directive.
This Lua code execution context does not support yielding, so Lua APIs that may yield (like cosockets and "light threads") are disabled in this context. One can usually work around this limitation by doing such operations in an earlier phase handler (like precontent_by_lua*) and passing along the result into this context via the ngx.ctx table.
This directive was first introduced in the v0.10.0 release.
syntax: balancer_by_lua_file <path-to-lua-script-file>
context: upstream
phase: content
Equivalent to balancer_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or, as from the v0.5.0rc32 release, the LuaJIT bytecode to be executed.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
This directive was first introduced in the v0.10.0 release.
syntax: balancer_keepalive <total-connections>
context: upstream
phase: loading-config
The total-connections parameter sets the maximum number of idle
keepalive connections to upstream servers that are preserved in the cache of
each worker process. When this number is exceeded, the least recently used
connections are closed.
It should be particularly noted that the keepalive directive does not limit the total number of connections to upstream servers that an nginx worker process can open. The connections parameter should be set to a number small enough to let upstream servers process new incoming connections as well.
This directive was first introduced in the v0.10.21 release.
syntax: lua_need_request_body <on|off>
default: off
context: http, server, location, location if
phase: depends on usage
Determines whether to force the request body data to be read before running rewrite/access/content_by_lua* or not. The Nginx core does not read the client request body by default and if request body data is required, then this directive should be turned on or the ngx.req.read_body function should be called within the Lua code.
To read the request body data within the $request_body variable, client_body_buffer_size must have the same value as client_max_body_size. Because when the content length exceeds client_body_buffer_size but less than client_max_body_size, Nginx will buffer the data into a temporary file on the disk, which will lead to empty value in the $request_body variable.
If the current location includes rewrite_by_lua* directives,
then the request body will be read just before the rewrite_by_lua* code is run (and also at the
rewrite phase). Similarly, if only content_by_lua is specified,
the request body will not be read until the content handler's Lua code is
about to run (i.e., the request body will be read during the content phase).
It is recommended however, to use the ngx.req.read_body and ngx.req.discard_body functions for finer control over the request body reading process instead.
This also applies to access_by_lua*.
syntax: ssl_client_hello_by_lua_block { lua-script }
context: http, server
phase: right-after-client-hello-message-was-processed
This directive runs user Lua code when Nginx is about to post-process the SSL client hello message for the downstream SSL (https) connections.
It is particularly useful for dynamically setting the SSL protocols according to the SNI.
It is also useful to do some custom operations according to the per-connection information in the client hello message.
For example, one can parse custom client hello extension and do the corresponding handling in pure Lua.
This Lua handler will always run whether the SSL session is resumed (via SSL session IDs or TLS session tickets) or not.
While the ssl_certificate_by_lua* Lua handler will only runs when initiating a full SSL handshake.
The ngx.ssl.clienthello Lua modules provided by the lua-resty-core library are particularly useful in this context.
Note that this handler runs in extremely early stage of SSL handshake, before the SSL client hello extensions are parsed.
So you can not use some Lua API like ssl.server_name() which is dependent on the later stage's processing.
Also note that only the directive in default server is valid for several virtual servers with the same IP address and port.
Below is a trivial example using the ngx.ssl.clienthello module at the same time:
server {
listen 443 ssl;
server_name test.com;
ssl_certificate /path/to/cert.crt;
ssl_certificate_key /path/to/key.key;
ssl_client_hello_by_lua_block {
local ssl_clt = require "ngx.ssl.clienthello"
local host, err = ssl_clt.get_client_hello_server_name()
if host == "test.com" then
ssl_clt.set_protocols({"TLSv1", "TLSv1.1"})
elseif host == "test2.com" then
ssl_clt.set_protocols({"TLSv1.2", "TLSv1.3"})
elseif not host then
ngx.log(ngx.ERR, "failed to get the SNI name: ", err)
ngx.exit(ngx.ERROR)
else
ngx.log(ngx.ERR, "unknown SNI name: ", host)
ngx.exit(ngx.ERROR)
end
}
...
}
server {
listen 443 ssl;
server_name test2.com;
ssl_certificate /path/to/cert.crt;
ssl_certificate_key /path/to/key.key;
...
}
See more information in the ngx.ssl.clienthello Lua modules' official documentation.
Uncaught Lua exceptions in the user Lua code immediately abort the current SSL session, so does the
ngx.exit call with an error code like ngx.ERROR.
This Lua code execution context does support yielding, so Lua APIs that may yield (like cosockets, sleeping, and "light threads") are enabled in this context
Note, you need to configure the ssl_certificate and ssl_certificate_key to avoid the following error while starting NGINX:
nginx: [emerg] no ssl configured for the server
This directive requires OpenSSL 1.1.1 or greater.
If you are using the official pre-built packages for OpenResty 1.21.4.1 or later, then everything should work out of the box.
If you are not using the Nginx core shipped with OpenResty 1.21.4.1 or later, you will need to apply patches to the standard Nginx core:
https://openresty.org/en/nginx-ssl-patches.html
Note for HTTP/3 (QUIC) users: When using this directive with HTTP/3 connections, certain yield operations may fail if the QUIC SSL Lua yield patch is not applied to your OpenSSL installation. OpenResty packages include this patch by default, but if you are building lua-nginx-module separately, you may need to apply the patch manually to ensure proper yield/resume functionality for HTTP/3 connections in SSL Lua phases. The patch can be found at: nginx-1.27.1-quic_ssl_lua_yield.patch
This directive was first introduced in the v0.10.21 release.
syntax: ssl_client_hello_by_lua_file <path-to-lua-script-file>
context: http, server
phase: right-after-client-hello-message-was-processed
Equivalent to ssl_client_hello_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or, as from the v0.5.0rc32 release, the LuaJIT bytecode to be executed.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
Note for HTTP/3 (QUIC) users: When using this directive with HTTP/3 connections, certain yield operations may fail if the QUIC SSL Lua yield patch is not applied to your OpenSSL installation. OpenResty packages include this patch by default, but if you are building lua-nginx-module separately, you may need to apply the patch manually to ensure proper yield/resume functionality for HTTP/3 connections in SSL Lua phases. The patch can be found at: nginx-1.27.1-quic_ssl_lua_yield.patch
This directive was first introduced in the v0.10.21 release.
syntax: ssl_certificate_by_lua_block { lua-script }
context: server
phase: right-before-SSL-handshake
This directive runs user Lua code when Nginx is about to start the SSL handshake for the downstream SSL (https) connections.
It is particularly useful for setting the SSL certificate chain and the corresponding private key on a per-request basis. It is also useful to load such handshake configurations nonblockingly from the remote (for example, with the cosocket API). And one can also do per-request OCSP stapling handling in pure Lua here as well.
Another typical use case is to do SSL handshake traffic control nonblockingly in this context, with the help of the lua-resty-limit-traffic#readme library, for example.
One can also do interesting things with the SSL handshake requests from the client side, like rejecting old SSL clients using the SSLv3 protocol or even below selectively.
The ngx.ssl and ngx.ocsp Lua modules provided by the lua-resty-core library are particularly useful in this context. You can use the Lua API offered by these two Lua modules to manipulate the SSL certificate chain and private key for the current SSL connection being initiated.
This Lua handler does not run at all, however, when Nginx/OpenSSL successfully resumes the SSL session via SSL session IDs or TLS session tickets for the current SSL connection. In other words, this Lua handler only runs when Nginx has to initiate a full SSL handshake.
Below is a trivial example using the ngx.ssl module at the same time:
server {
listen 443 ssl;
server_name test.com;
ssl_certificate_by_lua_block {
print("About to initiate a new SSL handshake!")
}
location / {
root html;
}
}
See more complicated examples in the ngx.ssl and ngx.ocsp Lua modules' official documentation.
Uncaught Lua exceptions in the user Lua code immediately abort the current SSL session, so does the
ngx.exit call with an error code like ngx.ERROR.
This Lua code execution context does support yielding, so Lua APIs that may yield (like cosockets, sleeping, and "light threads") are enabled in this context.
Note, however, you still need to configure the ssl_certificate and ssl_certificate_key directives even though you will not use this static certificate and private key at all. This is because the NGINX core requires their appearance otherwise you are seeing the following error while starting NGINX:
nginx: [emerg] no ssl configured for the server
This directive requires OpenSSL 1.0.2e or greater.
If you are using the official pre-built packages for OpenResty 1.9.7.2 or later, then everything should work out of the box.
If you are not using the Nginx core shipped with OpenResty 1.9.7.2 or later, you will need to apply patches to the standard Nginx core:
https://openresty.org/en/nginx-ssl-patches.html
Note for HTTP/3 (QUIC) users: When using this directive with HTTP/3 connections, certain yield operations may fail if the QUIC SSL Lua yield patch is not applied to your OpenSSL installation. OpenResty packages include this patch by default, but if you are building lua-nginx-module separately, you may need to apply the patch manually to ensure proper yield/resume functionality for HTTP/3 connections in SSL Lua phases. The patch can be found at: nginx-1.27.1-quic_ssl_lua_yield.patch
This directive was first introduced in the v0.10.0 release.
syntax: ssl_certificate_by_lua_file <path-to-lua-script-file>
context: server
phase: right-before-SSL-handshake
Equivalent to ssl_certificate_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or, as from the v0.5.0rc32 release, the LuaJIT bytecode to be executed.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
Note for HTTP/3 (QUIC) users: When using this directive with HTTP/3 connections, certain yield operations may fail if the QUIC SSL Lua yield patch is not applied to your OpenSSL installation. OpenResty packages include this patch by default, but if you are building lua-nginx-module separately, you may need to apply the patch manually to ensure proper yield/resume functionality for HTTP/3 connections in SSL Lua phases. The patch can be found at: nginx-1.27.1-quic_ssl_lua_yield.patch
This directive was first introduced in the v0.10.0 release.
syntax: ssl_session_fetch_by_lua_block { lua-script }
context: http
phase: right-before-SSL-handshake
This directive runs Lua code to look up and load the SSL session (if any) according to the session ID provided by the current SSL handshake request for the downstream.
The Lua API for obtaining the current session ID and loading a cached SSL session data is provided in the ngx.ssl.session Lua module shipped with the lua-resty-core library.
Lua APIs that may yield, like ngx.sleep and cosockets, are enabled in this context.
This hook, together with the ssl_session_store_by_lua* hook, can be used to implement distributed caching mechanisms in pure Lua (based on the cosocket API, for example). If a cached SSL session is found and loaded into the current SSL connection context, SSL session resumption can then get immediately initiated and bypass the full SSL handshake process which is very expensive in terms of CPU time.
Please note that TLS session tickets are very different and it is the clients' responsibility to cache the SSL session state when session tickets are used. SSL session resumptions based on TLS session tickets would happen automatically without going through this hook (nor the ssl_session_store_by_lua* hook). This hook is mainly for older or less capable SSL clients that can only do SSL sessions by session IDs.
When ssl_certificate_by_lua* is specified at the same time, this hook usually runs before ssl_certificate_by_lua*. When the SSL session is found and successfully loaded for the current SSL connection, SSL session resumption will happen and thus bypass the ssl_certificate_by_lua* hook completely. In this case, Nginx also bypasses the ssl_session_store_by_lua* hook, for obvious reasons.
To easily test this hook locally with a modern web browser, you can temporarily put the following line in your https server block to disable the TLS session ticket support:
ssl_session_tickets off;
But do not forget to comment this line out before publishing your site to the world.
If you are using the official pre-built packages for OpenResty 1.11.2.1 or later, then everything should work out of the box.
If you are not using one of the OpenSSL packages provided by OpenResty, you will need to apply patches to OpenSSL in order to use this directive:
https://openresty.org/en/openssl-patches.html
Similarly, if you are not using the Nginx core shipped with OpenResty 1.11.2.1 or later, you will need to apply patches to the standard Nginx core:
https://openresty.org/en/nginx-ssl-patches.html
This directive was first introduced in the v0.10.6 release.
Note that this directive can only be used in the http context starting
with the v0.10.7 release since SSL session resumption happens
before server name dispatch.
syntax: ssl_session_fetch_by_lua_file <path-to-lua-script-file>
context: http
phase: right-before-SSL-handshake
Equivalent to ssl_session_fetch_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or rather, the LuaJIT bytecode to be executed.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
This directive was first introduced in the v0.10.6 release.
Note that: this directive is only allowed to used in http context from the v0.10.7 release
(because SSL session resumption happens before server name dispatch).
syntax: ssl_session_store_by_lua_block { lua-script }
context: http
phase: right-after-SSL-handshake
This directive runs Lua code to fetch and save the SSL session (if any) according to the session ID provided by the current SSL handshake request for the downstream. The saved or cached SSL session data can be used for future SSL connections to resume SSL sessions without going through the full SSL handshake process (which is very expensive in terms of CPU time).
Lua APIs that may yield, like ngx.sleep and cosockets,
are disabled in this context. You can still, however, use the ngx.timer.at API
to create 0-delay timers to save the SSL session data asynchronously to external services (like redis or memcached).
The Lua API for obtaining the current session ID and the associated session state data is provided in the ngx.ssl.session Lua module shipped with the lua-resty-core library.
To easily test this hook locally with a modern web browser, you can temporarily put the following line in your https server block to disable the TLS session ticket support:
ssl_session_tickets off;
But do not forget to comment this line out before publishing your site to the world.
This directive was first introduced in the v0.10.6 release.
Note that: this directive is only allowed to used in http context from the v0.10.7 release
(because SSL session resumption happens before server name dispatch).
syntax: ssl_session_store_by_lua_file <path-to-lua-script-file>
context: http
phase: right-after-SSL-handshake
Equivalent to ssl_session_store_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or rather, the LuaJIT bytecode to be executed.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
This directive was first introduced in the v0.10.6 release.
Note that: this directive is only allowed to used in http context from the v0.10.7 release
(because SSL session resumption happens before server name dispatch).
syntax: proxy_ssl_certificate_by_lua_block { lua-script }
context: location
phase: right-after-server-certificate-request-message-was-processed
This directive runs user Lua code when Nginx is about to post-process the SSL server certificate request message from upstream. It is particularly useful for setting the SSL certificate chain and the corresponding private key for the upstream SSL (https) connections. It is also useful to load such handshake configurations nonblockingly from the remote (for example, with the cosocket API).
The ngx.ssl.proxysslcert Lua module provided by the lua-resty-core library are particularly useful in this context.
Below is a trivial example using the ngx.ssl module and the ngx.ssl.proxysslcert module at the same time:
server {
listen 443 ssl;
server_name test.com;
ssl_certificate /path/to/cert.crt;
ssl_certificate_key /path/to/key.key;
location /t {
proxy_pass https://upstream;
proxy_ssl_certificate_by_lua_block {
local ssl = require "ngx.ssl"
local proxy_ssl_cert = require "ngx.ssl.proxysslcert"
-- NOTE: for illustration only, we don't handle error below
local f = assert(io.open("/path/to/cert.crt"))
local cert_data = f:read("*a")
f:close()
local cert, err = ssl.parse_pem_cert(cert_data)
local ok, err = proxy_ssl_cert.set_cert(cert)
local f = assert(io.open("/path/to/key.key"))
local pkey_data = f:read("*a")
f:close()
local pkey, err = ssl.parse_pem_priv_key(pkey_data)
local ok, err = proxy_ssl_cert.set_priv_key(pkey)
-- ...
}
}
...
}
See more information in the ngx.ssl.proxysslcert Lua module's official documentation.
Uncaught Lua exceptions in the user Lua code immediately abort the current SSL session, so does the
ngx.exit call with an error code like ngx.ERROR.
This Lua code execution context does support yielding, so Lua APIs that may yield (like cosockets, sleeping, and "light threads") are enabled in this context.
Note that, unlike the relations between the ssl_certificate and ssl_certificate_key directives and ssl_certificate_by_lua*, the proxy_ssl_certificate and proxy_ssl_certificate_key directives can be used together with proxy_ssl_certificate_by_lua*.
When there are only proxy_ssl_certificate and proxy_ssl_certificate_key directives, the original Nginx behavior will obviously remain the same.
When there is only proxy_ssl_certificate_by_lua*, Nginx will send the certificate and its related private key and chain set by Lua codes.
When the proxy_ssl_certificate and proxy_ssl_certificate_key directives and proxy_ssl_certificate_by_lua* are used at the same time, then proxy_ssl_certificate_by_lua* will take precedence over the proxy_ssl_certificate and proxy_ssl_certificate_key directives.
Please refer to corresponding test case file and ngx.ssl.proxysslcert for more details.
Note also that, it has the same condition as the proxy_ssl_certificate directive for proxy_ssl_certificate_by_lua* to work, that is the upstream server should enable verification of client certificates.
This directive requires OpenSSL 1.0.2e or greater.
syntax: proxy_ssl_certificate_by_lua_file <path-to-lua-script-file>
context: location
phase: right-after-server-certificate-request-message-was-processed
Equivalent to proxy_ssl_certificate_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or, as from the v0.5.0rc32 release, the LuaJIT bytecode to be executed.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
syntax: proxy_ssl_verify_by_lua_block { lua-script }
context: location
phase: right-after-server-certificate-message-was-processed
This directive runs user Lua code when Nginx is about to post-process the SSL server certificate message for the upstream SSL (https) connections.
It is particularly useful to parse upstream server certificate and do some custom operations in pure lua.
The ngx.ssl.proxysslverify Lua module provided by the lua-resty-core library are particularly useful in this context.
Below is a trivial example using the ngx.ssl.proxysslverify module at the same time:
server {
listen 443 ssl;
server_name test.com;
ssl_certificate /path/to/cert.crt;
ssl_certificate_key /path/to/key.key;
location /t {
proxy_ssl_certificate /path/to/cert.crt;
proxy_ssl_certificate_key /path/to/key.key;
proxy_pass https://upstream;
proxy_ssl_verify_by_lua_block {
local proxy_ssl_vfy = require "ngx.ssl.proxysslverify"
local cert = proxy_ssl_vfy.get_verify_cert()
-- ocsp to verify cert
-- check crl
proxy_ssl_vfy.set_verify_result()
...
}
}
...
}
See more information in the ngx.ssl.proxysslverify Lua module's official documentation.
Uncaught Lua exceptions in the user Lua code immediately abort the current SSL session, so does the
ngx.exit call with an error code like ngx.ERROR.
This Lua code execution context does support yielding, so Lua APIs that may yield (like cosockets, sleeping, and "light threads") are enabled in this context
This directive requires OpenSSL 3.0.2 or greater.
syntax: proxy_ssl_verify_by_lua_file <path-to-lua-script-file>
context: location
phase: right-after-server-certificate-message-was-processed
Equivalent to proxy_ssl_verify_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or, as from the v0.5.0rc32 release, the LuaJIT bytecode to be executed.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
syntax: lua_shared_dict <name> <size>
default: no
context: http
phase: depends on usage
Declares a shared memory zone, <name>, to serve as storage for the shm based Lua dictionary ngx.shared.<name>.
Shared memory zones are always shared by all the Nginx worker processes in the current Nginx server instance.
The <size> argument accepts size units such as k and m:
http {
lua_shared_dict dogs 10m;
...
}
The hard-coded minimum size is 8KB while the practical minimum size depends on actual user data set (some people start with 12KB).
See ngx.shared.DICT for details.
This directive was first introduced in the v0.3.1rc22 release.
syntax: lua_socket_connect_timeout <time>
default: lua_socket_connect_timeout 60s
context: http, server, location
This directive controls the default timeout value used in TCP/unix-domain socket object's connect method and can be overridden by the settimeout or settimeouts methods.
The <time> argument can be an integer, with an optional time unit, like s (second), ms (millisecond), m (minute). The default time unit is s, i.e., "second". The default setting is 60s.
This directive was first introduced in the v0.5.0rc1 release.
syntax: lua_socket_send_timeout <time>
default: lua_socket_send_timeout 60s
context: http, server, location
Controls the default timeout value used in TCP/unix-domain socket object's send method and can be overridden by the settimeout or settimeouts methods.
The <time> argument can be an integer, with an optional time unit, like s (second), ms (millisecond), m (minute). The default time unit is s, i.e., "second". The default setting is 60s.
This directive was first introduced in the v0.5.0rc1 release.
syntax: lua_socket_send_lowat <size>
default: lua_socket_send_lowat 0
context: http, server, location
Controls the lowat (low water) value for the cosocket send buffer.
syntax: lua_socket_read_timeout <time>
default: lua_socket_read_timeout 60s
context: http, server, location
phase: depends on usage
This directive controls the default timeout value used in TCP/unix-domain socket object's receive method and iterator functions returned by the receiveuntil method. This setting can be overridden by the settimeout or settimeouts methods.
The <time> argument can be an integer, with an optional time unit, like s (second), ms (millisecond), m (minute). The default time unit is s, i.e., "second". The default setting is 60s.
This directive was first introduced in the v0.5.0rc1 release.
syntax: lua_socket_buffer_size <size>
default: lua_socket_buffer_size 4k/8k
context: http, server, location
Specifies the buffer size used by cosocket reading operations.
This buffer does not have to be that big to hold everything at the same time because cosocket supports 100% non-buffered reading and parsing. So even 1 byte buffer size should still work everywhere but the performance could be terrible.
This directive was first introduced in the v0.5.0rc1 release.
syntax: lua_socket_pool_size <size>
default: lua_socket_pool_size 30
context: http, server, location
Specifies the size limit (in terms of connection count) for every cosocket connection pool associated with every remote server (i.e., identified by either the host-port pair or the unix domain socket file path).
Default to 30 connections for every pool.
When the connection pool exceeds the available size limit, the least recently used (idle) connection already in the pool will be closed to make room for the current connection.
Note that the cosocket connection pool is per Nginx worker process rather than per Nginx server instance, so size limit specified here also applies to every single Nginx worker process.
This directive was first introduced in the v0.5.0rc1 release.
syntax: lua_socket_keepalive_timeout <time>
default: lua_socket_keepalive_timeout 60s
context: http, server, location
This directive controls the default maximal idle time of the connections in the cosocket built-in connection pool. When this timeout reaches, idle connections will be closed and removed from the pool. This setting can be overridden by cosocket objects' setkeepalive method.
The <time> argument can be an integer, with an optional time unit, like s (second), ms (millisecond), m (minute). The default time unit is s, i.e., "second". The default setting is 60s.
This directive was first introduced in the v0.5.0rc1 release.
syntax: lua_socket_log_errors on|off
default: lua_socket_log_errors on
context: http, server, location
This directive can be used to toggle error logging when a failure occurs for the TCP or UDP cosockets. If you are already doing proper error handling and logging in your Lua code, then it is recommended to turn this directive off to prevent data flushing in your Nginx error log files (which is usually rather expensive).
This directive was first introduced in the v0.5.13 release.
syntax: lua_ssl_ciphers <ciphers>
default: lua_ssl_ciphers DEFAULT
context: http, server, location
Specifies the enabled ciphers for requests to a SSL/TLS server in the tcpsock:sslhandshake method. The ciphers are specified in the format understood by the OpenSSL library.
The full list can be viewed using the “openssl ciphers” command.
This directive was first introduced in the v0.9.11 release.
syntax: lua_ssl_crl <file>
default: no
context: http, server, location
Specifies a file with revoked certificates (CRL) in the PEM format used to verify the certificate of the SSL/TLS server in the tcpsock:sslhandshake method.
This directive was first introduced in the v0.9.11 release.
syntax: lua_ssl_protocols [SSLv2] [SSLv3] [TLSv1] [TLSv1.1] [TLSv1.2] [TLSv1.3]
default: lua_ssl_protocols TLSv1 TLSv1.1 TLSv1.2 TLSv1.3
context: http, server, location
Enables the specified protocols for requests to a SSL/TLS server in the tcpsock:sslhandshake method.
The support for the TLSv1.3 parameter requires version v0.10.12 and OpenSSL 1.1.1.
From version v0.10.25, the default value change from SSLV3 TLSv1 TLSv1.1 TLSv1.2 to TLSv1 TLSv1.1 TLSv1.2 TLSv1.3.
This directive was first introduced in the v0.9.11 release.
syntax: lua_ssl_certificate <file>
default: none
context: http, server, location
Specifies the file path to the SSL/TLS certificate in PEM format used for the tcpsock:sslhandshake method.
This directive allows you to specify the SSL/TLS certificate that will be presented to server during the SSL/TLS handshake process.
This directive was first introduced in the v0.10.26 release.
See also lua_ssl_certificate_key and lua_ssl_verify_depth.
syntax: lua_ssl_certificate_key <file>
default: none
context: http, server, location
Specifies the file path to the private key associated with the SSL/TLS certificate used in the tcpsock:sslhandshake method.
This directive allows you to specify the private key file corresponding to the SSL/TLS certificate specified by lua_ssl_certificate. The private key should be in PEM format and must match the certificate.
This directive was first introduced in the v0.10.26 release.
See also lua_ssl_certificate and lua_ssl_verify_depth.
syntax: lua_ssl_trusted_certificate <file>
default: none
context: http, server, location
Specifies a file path with trusted CA certificates in the PEM format used to verify the certificate of the SSL/TLS server in the tcpsock:sslhandshake method.
This directive was first introduced in the v0.9.11 release.
See also lua_ssl_verify_depth.
syntax: lua_ssl_verify_depth <number>
default: lua_ssl_verify_depth 1
context: http, server, location
Sets the verification depth in the server certificates chain.
This directive was first introduced in the v0.9.11 release.
See also lua_ssl_certificate, lua_ssl_certificate_key and lua_ssl_trusted_certificate.
syntax: lua_ssl_key_log <file>
default: none
context: http, server, location
Enables logging of client connection SSL keys in the tcpsock:sslhandshake method and specifies the path to the key log file. Keys are logged in the SSLKEYLOGFILE format compatible with Wireshark.
syntax: lua_ssl_conf_command <command>
default: no
context: http, server, location
Sets arbitrary OpenSSL configuration commands.
The directive is supported when using OpenSSL 1.0.2 or higher and nginx 1.19.4 or higher. According to the specify command, higher OpenSSL version may be needed.
Several lua_ssl_conf_command directives can be specified on the same level:
lua_ssl_conf_command Options PrioritizeChaCha;
lua_ssl_conf_command Ciphersuites TLS_CHACHA20_POLY1305_SHA256;
Configuration commands are applied after OpenResty own configuration for SSL, so they can be used to override anything set by OpenResty.
Note though that configuring OpenSSL directly with lua_ssl_conf_command might result in a behaviour OpenResty does not expect, and should be done with care.
This directive was first introduced in the v0.10.21 release.
syntax: lua_upstream_skip_openssl_default_verify on|off
default: lua_upstream_skip_openssl_default_verify off
context: location, location-if
When using proxy_ssl_verify_by_lua directive, lua_upstream_skip_openssl_default_verify controls whether to skip default openssl's verify function, that means using pure Lua code to verify upstream server certificate.
This directive is turned off by default.
syntax: lua_http10_buffering on|off
default: lua_http10_buffering on
context: http, server, location, location-if
Enables or disables automatic response buffering for HTTP 1.0 (or older) requests. This buffering mechanism is mainly used for HTTP 1.0 keep-alive which relies on a proper Content-Length response header.
If the Lua code explicitly sets a Content-Length response header before sending the headers (either explicitly via ngx.send_headers or implicitly via the first ngx.say or ngx.print call), then the HTTP 1.0 response buffering will be disabled even when this directive is turned on.
To output very large response data in a streaming fashion (via the ngx.flush call, for example), this directive MUST be turned off to minimize memory usage.
This directive is turned on by default.
This directive was first introduced in the v0.5.0rc19 release.
syntax: rewrite_by_lua_no_postpone on|off
default: rewrite_by_lua_no_postpone off
context: http
Controls whether or not to disable postponing rewrite_by_lua* directives to run at the end of the rewrite request-processing phase. By default, this directive is turned off and the Lua code is postponed to run at the end of the rewrite phase.
This directive was first introduced in the v0.5.0rc29 release.
syntax: access_by_lua_no_postpone on|off
default: access_by_lua_no_postpone off
context: http
Controls whether or not to disable postponing access_by_lua* directives to run at the end of the access request-processing phase. By default, this directive is turned off and the Lua code is postponed to run at the end of the access phase.
This directive was first introduced in the v0.9.20 release.
syntax: precontent_by_lua_no_postpone on|off
default: precontent_by_lua_no_postpone off
context: http
Controls whether or not to disable postponing precontent_by_lua* directives to run at the end of the precontent request-processing phase. By default, this directive is turned off and the Lua code is postponed to run at the end of the precontent phase.
syntax: lua_transform_underscores_in_response_headers on|off
default: lua_transform_underscores_in_response_headers on
context: http, server, location, location-if
Controls whether to transform underscores (_) in the response header names specified in the ngx.header.HEADER API to hyphens (-).
This directive was first introduced in the v0.5.0rc32 release.
syntax: lua_check_client_abort on|off
default: lua_check_client_abort off
context: http, server, location, location-if
This directive controls whether to check for premature client connection abortion.
When this directive is on, the ngx_lua module will monitor the premature connection close event on the downstream connections and when there is such an event, it will call the user Lua function callback (registered by ngx.on_abort) or just stop and clean up all the Lua "light threads" running in the current request's request handler when there is no user callback function registered.
According to the current implementation, however, if the client closes the connection before the Lua code finishes reading the request body data via ngx.req.socket, then ngx_lua will neither stop all the running "light threads" nor call the user callback (if ngx.on_abort has been called). Instead, the reading operation on ngx.req.socket will just return the error message "client aborted" as the second return value (the first return value is surely nil).
When TCP keepalive is disabled, it is relying on the client side to close the socket gracefully (by sending a FIN packet or something like that). For (soft) real-time web applications, it is highly recommended to configure the TCP keepalive support in your system's TCP stack implementation in order to detect "half-open" TCP connections in time.
For example, on Linux, you can configure the standard listen directive in your nginx.conf file like this:
listen 80 so_keepalive=2s:2s:8;
On FreeBSD, you can only tune the system-wide configuration for TCP keepalive, for example:
# sysctl net.inet.tcp.keepintvl=2000
# sysctl net.inet.tcp.keepidle=2000
This directive was first introduced in the v0.7.4 release.
See also ngx.on_abort.
syntax: lua_max_pending_timers <count>
default: lua_max_pending_timers 1024
context: http
Controls the maximum number of pending timers allowed.
Pending timers are those timers that have not expired yet.
When exceeding this limit, the ngx.timer.at call will immediately return nil and the error string "too many pending timers".
This directive was first introduced in the v0.8.0 release.
syntax: lua_max_running_timers <count>
default: lua_max_running_timers 256
context: http
Controls the maximum number of "running timers" allowed.
Running timers are those timers whose user callback functions are still running or lightthreads spawned in callback functions are still running.
When exceeding this limit, Nginx will stop running the callbacks of newly expired timers and log an error message "N lua_max_running_timers are not enough" where "N" is the current value of this directive.
This directive was first introduced in the v0.8.0 release.
syntax: lua_sa_restart on|off
default: lua_sa_restart on
context: http
When enabled, this module will set the SA_RESTART flag on Nginx workers signal dispositions.
This allows Lua I/O primitives to not be interrupted by Nginx's handling of various signals.
This directive was first introduced in the v0.10.14 release.
syntax: lua_worker_thread_vm_pool_size <size>
default: lua_worker_thread_vm_pool_size 10
context: http
Specifies the size limit of the Lua VM pool (default 100) that will be used in the ngx.run_worker_thread API.
Also, it is not allowed to create Lua VMs that exceeds the pool size limit.
The Lua VM in the VM pool is used to execute Lua code in separate thread.
The pool is global at Nginx worker level. And it is used to reuse Lua VMs between requests.
Warning: Each worker thread uses a separate Lua VM and caches the Lua VM for reuse in subsequent operations. Configuring too many worker threads can result in consuming a lot of memory.
The various *_by_lua, *_by_lua_block and *_by_lua_file configuration directives serve as gateways to the Lua API within the nginx.conf file. The Nginx Lua API described below can only be called within the user Lua code run in the context of these configuration directives.
The API is exposed to Lua in the form of two standard packages ngx and ndk. These packages are in the default global scope within ngx_lua and are always available within ngx_lua directives.
The packages can be introduced into external Lua modules like this:
local say = ngx.say
local _M = {}
function _M.foo(a)
say(a)
end
return _M
Use of the package.seeall flag is strongly discouraged due to its various bad side-effects.
It is also possible to directly require the packages in external Lua modules:
local ngx = require "ngx"
local ndk = require "ndk"
The ability to require these packages was introduced in the v0.2.1rc19 release.
Network I/O operations in user code should only be done through the Nginx Lua API calls as the Nginx event loop may be blocked and performance drop off dramatically otherwise. Disk operations with relatively small amount of data can be done using the standard Lua io library but huge file reading and writing should be avoided wherever possible as they may block the Nginx process significantly. Delegating all network and disk I/O operations to Nginx's subrequests (via the ngx.location.capture method and similar) is strongly recommended for maximum performance.
syntax: val = ngx.arg[index]
context: set_by_lua*, body_filter_by_lua*
When this is used in the context of the set_by_lua* directives, this table is read-only and holds the input arguments to the config directives:
value = ngx.arg[n]
Here is an example
location /foo {
set $a 32;
set $b 56;
set_by_lua $sum
'return tonumber(ngx.arg[1]) + tonumber(ngx.arg[2])'
$a $b;
echo $sum;
}
that writes out 88, the sum of 32 and 56.
When this table is used in the context of body_filter_by_lua*, the first element holds the input data chunk to the output filter code and the second element holds the boolean flag for the "eof" flag indicating the end of the whole output data stream.
The data chunk and "eof" flag passed to the downstream Nginx output filters can also be overridden by assigning values directly to the corresponding table elements. When setting nil or an empty Lua string value to ngx.arg[1], no data chunk will be passed to the downstream Nginx output filters at all.
syntax: ngx.var.VAR_NAME
context: set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*, body_filter_by_lua*, log_by_lua*, balancer_by_lua*
Read and write Nginx variable values.
value = ngx.var.some_nginx_variable_name
ngx.var.some_nginx_variable_name = value
Note that only already defined Nginx variables can be written to. For example:
location /foo {
set $my_var ''; # this line is required to create $my_var at config time
content_by_lua_block {
ngx.var.my_var = 123
...
}
}
That is, Nginx variables cannot be created on-the-fly. Here is a list of pre-defined Nginx variables.
Some special Nginx variables like $args and $limit_rate can be assigned a value,
many others are not, like $query_string, $arg_PARAMETER, and $http_NAME.
Nginx regex group capturing variables $1, $2, $3, and etc, can be read by this
interface as well, by writing ngx.var[1], ngx.var[2], ngx.var[3], and etc.
Setting ngx.var.Foo to a nil value will unset the $Foo Nginx variable.
ngx.var.args = nil
CAUTION When reading from an Nginx variable, Nginx will allocate memory in the per-request memory pool which is freed only at request termination. So when you need to read from an Nginx variable repeatedly in your Lua code, cache the Nginx variable value to your own Lua variable, for example,
local val = ngx.var.some_var
--- use the val repeatedly later
to prevent (temporary) memory leaking within the current request's lifetime. Another way of caching the result is to use the ngx.ctx table.
Undefined Nginx variables are evaluated to nil while uninitialized (but defined) Nginx variables are evaluated to an empty Lua string.
This API requires a relatively expensive metamethod call and it is recommended to avoid using it on hot code paths.
context: init_by_lua*, set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*, body_filter_by_lua*, *log_by_lua*, ngx.timer.*, balancer_by_lua*, ssl_certificate_by_lua*, ssl_session_fetch_by_lua*, ssl_session_store_by_lua*, ssl_client_hello_by_lua*
ngx.OK (0)
ngx.ERROR (-1)
ngx.AGAIN (-2)
ngx.DONE (-4)
ngx.DECLINED (-5)
Note that only three of these constants are utilized by the Nginx API for Lua (i.e., ngx.exit accepts ngx.OK, ngx.ERROR, and ngx.DECLINED as input).
ngx.null
The ngx.null constant is a NULL light userdata usually used to represent nil values in Lua tables etc and is similar to the lua-cjson library's cjson.null constant. This constant was first introduced in the v0.5.0rc5 release.
The ngx.DECLINED constant was first introduced in the v0.5.0rc19 release.
context: init_by_lua*, set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*, body_filter_by_lua*, log_by_lua*, ngx.timer.*, balancer_by_lua*, ssl_certificate_by_lua*, ssl_session_fetch_by_lua*, ssl_session_store_by_lua*, ssl_client_hello_by_lua*
ngx.HTTP_GET
ngx.HTTP_HEAD
ngx.HTTP_PUT
ngx.HTTP_POST
ngx.HTTP_DELETE
ngx.HTTP_OPTIONS (added in the v0.5.0rc24 release)
ngx.HTTP_MKCOL (added in the v0.8.2 release)
ngx.HTTP_COPY (added in the v0.8.2 release)
ngx.HTTP_MOVE (added in the v0.8.2 release)
ngx.HTTP_PROPFIND (added in the v0.8.2 release)
ngx.HTTP_PROPPATCH (added in the v0.8.2 release)
ngx.HTTP_LOCK (added in the v0.8.2 release)
ngx.HTTP_UNLOCK (added in the v0.8.2 release)
ngx.HTTP_PATCH (added in the v0.8.2 release)
ngx.HTTP_TRACE (added in the v0.8.2 release)
These constants are usually used in ngx.location.capture and ngx.location.capture_multi method calls.
context: init_by_lua*, set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*, body_filter_by_lua*, log_by_lua*, ngx.timer.*, balancer_by_lua*, ssl_certificate_by_lua*, ssl_session_fetch_by_lua*, ssl_session_store_by_lua*, ssl_client_hello_by_lua*
value = ngx.HTTP_CONTINUE (100) (first added in the v0.9.20 release)
value = ngx.HTTP_SWITCHING_PROTOCOLS (101) (first added in the v0.9.20 release)
value = ngx.HTTP_OK (200)
value = ngx.HTTP_CREATED (201)
value = ngx.HTTP_ACCEPTED (202) (first added in the v0.9.20 release)
value = ngx.HTTP_NO_CONTENT (204) (first added in the v0.9.20 release)
value = ngx.HTTP_PARTIAL_CONTENT (206) (first added in the v0.9.20 release)
value = ngx.HTTP_SPECIAL_RESPONSE (300)
value = ngx.HTTP_MOVED_PERMANENTLY (301)
value = ngx.HTTP_MOVED_TEMPORARILY (302)
value = ngx.HTTP_SEE_OTHER (303)
value = ngx.HTTP_NOT_MODIFIED (304)
value = ngx.HTTP_TEMPORARY_REDIRECT (307) (first added in the v0.9.20 release)
value = ngx.HTTP_PERMANENT_REDIRECT (308)
value = ngx.HTTP_BAD_REQUEST (400)
value = ngx.HTTP_UNAUTHORIZED (401)
value = ngx.HTTP_PAYMENT_REQUIRED (402) (first added in the v0.9.20 release)
value = ngx.HTTP_FORBIDDEN (403)
value = ngx.HTTP_NOT_FOUND (404)
value = ngx.HTTP_NOT_ALLOWED (405)
value = ngx.HTTP_NOT_ACCEPTABLE (406) (first added in the v0.9.20 release)
value = ngx.HTTP_REQUEST_TIMEOUT (408) (first added in the v0.9.20 release)
value = ngx.HTTP_CONFLICT (409) (first added in the v0.9.20 release)
value = ngx.HTTP_GONE (410)
value = ngx.HTTP_UPGRADE_REQUIRED (426) (first added in the v0.9.20 release)
value = ngx.HTTP_TOO_MANY_REQUESTS (429) (first added in the v0.9.20 release)
value = ngx.HTTP_CLOSE (444) (first added in the v0.9.20 release)
value = ngx.HTTP_ILLEGAL (451) (first added in the v0.9.20 release)
value = ngx.HTTP_INTERNAL_SERVER_ERROR (500)
value = ngx.HTTP_NOT_IMPLEMENTED (501)
value = ngx.HTTP_METHOD_NOT_IMPLEMENTED (501) (kept for compatibility)
value = ngx.HTTP_BAD_GATEWAY (502) (first added in the v0.9.20 release)
value = ngx.HTTP_SERVICE_UNAVAILABLE (503)
value = ngx.HTTP_GATEWAY_TIMEOUT (504) (first added in the v0.3.1rc38 release)
value = ngx.HTTP_VERSION_NOT_SUPPORTED (505) (first added in the v0.9.20 release)
value = ngx.HTTP_INSUFFICIENT_STORAGE (507) (first added in the v0.9.20 release)
context: init_by_lua*, init_worker_by_lua*, set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*, body_filter_by_lua*, log_by_lua*, ngx.timer.*, balancer_by_lua*, ssl_certificate_by_lua*, ssl_session_fetch_by_lua*, ssl_session_store_by_lua*, exit_worker_by_lua*, ssl_client_hello_by_lua*
ngx.STDERR
ngx.EMERG
ngx.ALERT
ngx.CRIT
ngx.ERR
ngx.WARN
ngx.NOTICE
ngx.INFO
ngx.DEBUG
These constants are usually used by the ngx.log method.
syntax: print(...)
context: init_by_lua*, init_worker_by_lua*, set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*, body_filter_by_lua*, log_by_lua*, ngx.timer.*, balancer_by_lua*, ssl_certificate_by_lua*, ssl_session_fetch_by_lua*, ssl_session_store_by_lua*, exit_worker_by_lua*, ssl_client_hello_by_lua*
Writes argument values into the Nginx error.log file with the ngx.NOTICE log level.
It is equivalent to
ngx.log(ngx.NOTICE, ...)
Lua nil arguments are accepted and result in literal "nil" strings while Lua booleans result in literal "true" or "false" strings. And the ngx.null constant will yield the "null" string output.
There is a hard coded 2048 byte limitation on error message lengths in the Nginx core. This limit includes trailing newlines and leading time stamps. If the message size exceeds this limit, Nginx will truncate the message text accordingly. This limit can be manually modified by editing the NGX_MAX_ERROR_STR macro definition in the src/core/ngx_log.h file in the Nginx source tree.
context: init_worker_by_lua*, set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*, body_filter_by_lua*, log_by_lua*, ngx.timer.*, balancer_by_lua*, exit_worker_by_lua*
This table can be used to store per-request Lua context data and has a life time identical to the current request (as with the Nginx variables).
Consider the following example,
location /test {
rewrite_by_lua_block {
ngx.ctx.foo = 76
}
access_by_lua_block {
ngx.ctx.foo = ngx.ctx.foo + 3
}
content_by_lua_block {
ngx.say(ngx.ctx.foo)
}
}
Then GET /test will yield the output
79
That is, the ngx.ctx.foo entry persists across the rewrite, access, and content phases of a request.
Every request, including subrequests, has its own copy of the table. For example:
location /sub {
content_by_lua_block {
ngx.say("sub pre: ", ngx.ctx.blah)
ngx.ctx.blah = 32
ngx.say("sub post: ", ngx.ctx.blah)
}
}
location /main {
content_by_lua_block {
ngx.ctx.blah = 73
ngx.say("main pre: ", ngx.ctx.blah)
local res = ngx.location.capture("/sub")
ngx.print(res.body)
ngx.say("main post: ", ngx.ctx.blah)
}
}
Then GET /main will give the output
main pre: 73
sub pre: nil
sub post: 32
main post: 73
Here, modification of the ngx.ctx.blah entry in the subrequest does not affect the one in the parent request. This is because they have two separate versions of ngx.ctx.blah.
Internal redirects (triggered by nginx configuration directives like error_page, try_files, index, etc.) will destroy the original request ngx.ctx data (if any) and the new request will have an empty ngx.ctx table. For instance,
location /new {
content_by_lua_block {
ngx.say(ngx.ctx.foo)
}
}
location /orig {
content_by_lua_block {
ngx.ctx.foo = "hello"
ngx.exec("/new")
}
}
Then GET /orig will give
nil
rather than the original "hello" value.
Because HTTP request is created after SSL handshake, the ngx.ctx created
in ssl_certificate_by_lua*, ssl_session_store_by_lua*, ssl_session_fetch_by_lua* and ssl_client_hello_by_lua*
is not available in the following phases like rewrite_by_lua*.
Since v0.10.18, the ngx.ctx created during a SSL handshake
will be inherited by the requests which share the same TCP connection established by the handshake.
Note that overwrite values in ngx.ctx in the http request phases (like rewrite_by_lua*) will only take affect in the current http request.
Arbitrary data values, including Lua closures and nested tables, can be inserted into this "magic" table. It also allows the registration of custom meta methods.
Overriding ngx.ctx with a new Lua table is also supported, for example,
ngx.ctx = { foo = 32, bar = 54 }
When being used in the context of init_worker_by_lua*, this table just has the same lifetime of the current Lua handler.
The ngx.ctx lookup requires relatively expensive metamethod calls and it is much slower than explicitly passing per-request data along by your own function arguments. So do not abuse this API for saving your own function arguments because it usually has quite some performance impact.
Because of the metamethod magic, never "local" the ngx.ctx table outside your Lua function scope on the Lua module level due to worker-level data sharing. For example, the following is bad:
-- mymodule.lua
local _M = {}
-- the following line is bad since ngx.ctx is a per-request
-- data while this <code>ctx</code> variable is on the Lua module level
-- and thus is per-nginx-worker.
local ctx = ngx.ctx
function _M.main()
ctx.foo = "bar"
end
return _M
Use the following instead:
-- mymodule.lua
local _M = {}
function _M.main(ctx)
ctx.foo = "bar"
end
return _M
That is, let the caller pass the ctx table explicitly via a function argument.
syntax: res = ngx.location.capture(uri, options?)
context: rewrite_by_lua*, access_by_lua*, content_by_lua*
Issues a synchronous but still non-blocking Nginx Subrequest using uri.
Nginx's subrequests provide a powerful way to make non-blocking internal requests to other locations configured with disk file directory or any other Nginx C modules like ngx_proxy, ngx_fastcgi, ngx_memc,
ngx_postgres, ngx_drizzle, and even ngx_lua itself and etc etc etc.
Also note that subrequests just mimic the HTTP interface but there is no extra HTTP/TCP traffic nor IPC involved. Everything works internally, efficiently, on the C level.
Subrequests are completely different from HTTP 301/302 redirection (via ngx.redirect) and internal redirection (via ngx.exec).
You should always read the request body (by either calling ngx.req.read_body or configuring lua_need_request_body on) before initiating a subrequest.
This API function (as well as ngx.location.capture_multi) always buffers the whole response body of the subrequest in memory. Thus, you should use cosockets and streaming processing instead if you have to handle large subrequest responses.
Here is a basic example:
res = ngx.location.capture(uri)
Returns a Lua table with 4 slots: res.status, res.header, res.body, and res.truncated.
res.status holds the response status code for the subrequest response.
res.header holds all the response headers of the
subrequest and it is a normal Lua table. For multi-value response headers,
the value is a Lua (array) table that holds all the values in the order that
they appear. For instance, if the subrequest response headers contain the following
lines:
Set-Cookie: a=3
Set-Cookie: foo=bar
Set-Cookie: baz=blah
Then res.header["Set-Cookie"] will be evaluated to the table value
{"a=3", "foo=bar", "baz=blah"}.
res.body holds the subrequest's response body data, which might be truncated. You always need to check the res.truncated boolean flag to see if res.body contains truncated data. The data truncation here can only be caused by those unrecoverable errors in your subrequests like the cases that the remote end aborts the connection prematurely in the middle of the response body data stream or a read timeout happens when your subrequest is receiving the response body data from the remote.
URI query strings can be concatenated to URI itself, for instance,
res = ngx.location.capture('/foo/bar?a=3&b=4')
Named locations like @foo are not allowed due to a limitation in
the Nginx core. Use normal locations combined with the internal directive to
prepare internal-only locations.
An optional option table can be fed as the second argument, which supports the options:
method
specify the subrequest's request method, which only accepts constants like ngx.HTTP_POST.body
specify the subrequest's request body (string value only).args
specify the subrequest's URI query arguments (both string value and Lua tables are accepted)headers
specify the subrequest's request headers (Lua table only). this headers will override the original headers of the subrequest.ctx
specify a Lua table to be the ngx.ctx table for the subrequest. It can be the current request's ngx.ctx table, which effectively makes the parent and its subrequest to share exactly the same context table. This option was first introduced in the v0.3.1rc25 release.vars
take a Lua table which holds the values to set the specified Nginx variables in the subrequest as this option's value. This option was first introduced in the v0.3.1rc31 release.copy_all_vars
specify whether to copy over all the Nginx variable values of the current request to the subrequest in question. modifications of the Nginx variables in the subrequest will not affect the current (parent) request. This option was first introduced in the v0.3.1rc31 release.share_all_vars
specify whether to share all the Nginx variables of the subrequest with the current (parent) request. modifications of the Nginx variables in the subrequest will affect the current (parent) request. Enabling this option may lead to hard-to-debug issues due to bad side-effects and is considered bad and harmful. Only enable this option when you completely know what you are doing.always_forward_body
when set to true, the current (parent) request's request body will always be forwarded to the subrequest being created if the body option is not specified. The request body read by either ngx.req.read_body() or lua_need_request_body on will be directly forwarded to the subrequest without copying the whole request body data when creating the subrequest (no matter the request body data is buffered in memory buffers or temporary files). By default, this option is false and when the body option is not specified, the request body of the current (parent) request is only forwarded when the subrequest takes the PUT or POST request method.Issuing a POST subrequest, for example, can be done as follows
res = ngx.location.capture(
'/foo/bar',
{ method = ngx.HTTP_POST, body = 'hello, world' }
)
See HTTP method constants methods other than POST.
The method option is ngx.HTTP_GET by default.
The args option can specify extra URI arguments, for instance,
ngx.location.capture('/foo?a=1',
{ args = { b = 3, c = ':' } }
)
is equivalent to
ngx.location.capture('/foo?a=1&b=3&c=%3a')
that is, this method will escape argument keys and values according to URI rules and
concatenate them together into a complete query string. The format for the Lua table passed as the args argument is identical to the format used in the ngx.encode_args method.
The args option can also take plain query strings:
ngx.location.capture('/foo?a=1',
{ args = 'b=3&c=%3a' }
)
This is functionally identical to the previous examples.
The share_all_vars option controls whether to share Nginx variables among the current request and its subrequests.
If this option is set to true, then the current request and associated subrequests will share the same Nginx variable scope. Hence, changes to Nginx variables made by a subrequest will affect the current request.
Care should be taken in using this option as variable scope sharing can have unexpected side effects. The args, vars, or copy_all_vars options are generally preferable instead.
This option is set to false by default
location /other {
set $dog "$dog world";
echo "$uri dog: $dog";
}
location /lua {
set $dog 'hello';
content_by_lua_block {
res = ngx.location.capture("/other",
{ share_all_vars = true })
ngx.print(res.body)
ngx.say(ngx.var.uri, ": ", ngx.var.dog)
}
}
Accessing location /lua gives
/other dog: hello world
/lua: hello world
The copy_all_vars option provides a copy of the parent request's Nginx variables to subrequests when such subrequests are issued. Changes made to these variables by such subrequests will not affect the parent request or any other subrequests sharing the parent request's variables.
location /other {
set $dog "$dog world";
echo "$uri dog: $dog";
}
location /lua {
set $dog 'hello';
content_by_lua_block {
res = ngx.location.capture("/other",
{ copy_all_vars = true })
ngx.print(res.body)
ngx.say(ngx.var.uri, ": ", ngx.var.dog)
}
}
Request GET /lua will give the output
/other dog: hello world
/lua: hello
Note that if both share_all_vars and copy_all_vars are set to true, then share_all_vars takes precedence.
In addition to the two settings above, it is possible to specify
values for variables in the subrequest using the vars option. These
variables are set after the sharing or copying of variables has been
evaluated, and provides a more efficient method of passing specific
values to a subrequest over encoding them as URL arguments and
unescaping them in the Nginx config file.
location /other {
content_by_lua_block {
ngx.say("dog = ", ngx.var.dog)
ngx.say("cat = ", ngx.var.cat)
}
}
location /lua {
set $dog '';
set $cat '';
content_by_lua_block {
res = ngx.location.capture("/other",
{ vars = { dog = "hello", cat = 32 }})
ngx.print(res.body)
}
}
Accessing /lua will yield the output
dog = hello
cat = 32
The headers option can be used to specify the request headers for the subrequest. The value of this option should be a Lua table where the keys are the header names and the values are the header values. For example,
location /foo {
content_by_lua_block {
ngx.print(ngx.var.http_x_test)
}
}
location /lua {
content_by_lua_block {
local res = ngx.location.capture("/foo", {
headers = {
["X-Test"] = "aa",
}
})
ngx.print(res.body)
}
}
Accessing /lua will yield the output
aa
The ctx option can be used to specify a custom Lua table to serve as the ngx.ctx table for the subrequest.
location /sub {
content_by_lua_block {
ngx.ctx.foo = "bar";
}
}
location /lua {
content_by_lua_block {
local ctx = {}
res = ngx.location.capture("/sub", { ctx = ctx })
ngx.say(ctx.foo)
ngx.say(ngx.ctx.foo)
}
}
Then request GET /lua gives
bar
nil
It is also possible to use this ctx option to share the same ngx.ctx table between the current (parent) request and the subrequest:
location /sub {
content_by_lua_block {
ngx.ctx.foo = "bar"
}
}
location /lua {
content_by_lua_block {
res = ngx.location.capture("/sub", { ctx = ngx.ctx })
ngx.say(ngx.ctx.foo)
}
}
Request GET /lua yields the output
bar
Note that subrequests issued by ngx.location.capture inherit all the
request headers of the current request by default and that this may have unexpected side effects on the
subrequest responses. For example, when using the standard ngx_proxy module to serve
subrequests, an "Accept-Encoding: gzip" header in the main request may result
in gzipped responses that cannot be handled properly in Lua code. Original request headers should be ignored by setting
proxy_pass_request_headers to off in subrequest locations.
When the body option is not specified and the always_forward_body option is false (the default value), the POST and PUT subrequests will inherit the request bodies of the parent request (if any).
There is a hard-coded upper limit on the number of subrequests possible for every main request. In older versions of Nginx, the limit was 50 concurrent subrequests and in more recent versions, Nginx 1.9.5 onwards, the same limit is changed to limit the depth of recursive subrequests. When this limit is exceeded, the following error message is added to the error.log file:
[error] 13983#0: *1 subrequests cycle while processing "/uri"
The limit can be manually modified if required by editing the definition of the NGX_HTTP_MAX_SUBREQUESTS macro in the nginx/src/http/ngx_http_request.h file in the Nginx source tree.
Please also refer to restrictions on capturing locations configured by subrequest directives of other modules.
syntax: res1, res2, ... = ngx.location.capture_multi({ {uri, options?}, {uri, options?}, ... })
context: rewrite_by_lua*, access_by_lua*, content_by_lua*
Just like ngx.location.capture, but supports multiple subrequests running in parallel.
This function issues several parallel subrequests specified by the input table and returns their results in the same order. For example,
res1, res2, res3 = ngx.location.capture_multi{
{ "/foo", { args = "a=3&b=4" } },
{ "/bar" },
{ "/baz", { method = ngx.HTTP_POST, body = "hello" } },
}
if res1.status == ngx.HTTP_OK then
...
end
if res2.body == "BLAH" then
...
end
This function will not return until all the subrequests terminate. The total latency is the longest latency of the individual subrequests rather than the sum.
Lua tables can be used for both requests and responses when the number of subrequests to be issued is not known in advance:
-- construct the requests table
local reqs = {}
table.insert(reqs, { "/mysql" })
table.insert(reqs, { "/postgres" })
table.insert(reqs, { "/redis" })
table.insert(reqs, { "/memcached" })
-- issue all the requests at once and w
Truncated — view the full README on GitHub.
(top 30 of 150)
C
94.8%
Lua
3.2%
Shell
1.0%
Embed the Power of Lua into NGINX HTTP servers
11,788
stars
3,973
commits
C
primary language
Sep 7, 2026
updated
ngx_http_lua_module - Embed the power of Lua into Nginx HTTP Servers.
This module is a core component of OpenResty. If you are using this module, then you are essentially using OpenResty :)
This module is not distributed with the Nginx source. See the installation instructions.
Production ready.
This document describes ngx_lua v0.10.29, which was released on Oct 24, 2025.
YouTube video "Hello World HTTP Example with OpenResty/Lua"
YouTube video "Write Your Own Lua Modules in OpenResty/Nginx Applications"
YouTube video "OpenResty's resty Command-Line Utility Demo"
YouTube video "Measure Execution Time of Lua Code Correctly in OpenResty"
YouTube video "Precompile Lua Modules into LuaJIT Bytecode to Speedup OpenResty Startup"
You are welcome to subscribe to our official YouTube channel, OpenResty.
# set search paths for pure Lua external libraries (';;' is the default path):
lua_package_path '/foo/bar/?.lua;/blah/?.lua;;';
# set search paths for Lua external libraries written in C (can also use ';;'):
lua_package_cpath '/bar/baz/?.so;/blah/blah/?.so;;';
server {
location /lua_content {
# MIME type determined by default_type:
default_type 'text/plain';
content_by_lua_block {
ngx.say('Hello,world!')
}
}
location /nginx_var {
# MIME type determined by default_type:
default_type 'text/plain';
# try access /nginx_var?a=hello,world
content_by_lua_block {
ngx.say(ngx.var.arg_a)
}
}
location = /request_body {
client_max_body_size 50k;
client_body_buffer_size 50k;
content_by_lua_block {
ngx.req.read_body() -- explicitly read the req body
local data = ngx.req.get_body_data()
if data then
ngx.say("body data:")
ngx.print(data)
return
end
-- body may get buffered in a temp file:
local file = ngx.req.get_body_file()
if file then
ngx.say("body is in file ", file)
else
ngx.say("no body found")
end
}
}
# transparent non-blocking I/O in Lua via subrequests
# (well, a better way is to use cosockets)
location = /lua {
# MIME type determined by default_type:
default_type 'text/plain';
content_by_lua_block {
local res = ngx.location.capture("/some_other_location")
if res then
ngx.say("status: ", res.status)
ngx.say("body:")
ngx.print(res.body)
end
}
}
location = /foo {
rewrite_by_lua_block {
res = ngx.location.capture("/memc",
{ args = { cmd = "incr", key = ngx.var.uri } }
)
}
proxy_pass http://blah.blah.com;
}
location = /mixed {
rewrite_by_lua_file /path/to/rewrite.lua;
access_by_lua_file /path/to/access.lua;
content_by_lua_file /path/to/content.lua;
}
# use nginx var in code path
# CAUTION: contents in nginx var must be carefully filtered,
# otherwise there'll be great security risk!
location ~ ^/app/([-_a-zA-Z0-9/]+) {
set $path $1;
content_by_lua_file /path/to/lua/app/root/$path.lua;
}
location / {
client_max_body_size 100k;
client_body_buffer_size 100k;
access_by_lua_block {
-- check the client IP address is in our black list
if ngx.var.remote_addr == "132.5.72.3" then
ngx.exit(ngx.HTTP_FORBIDDEN)
end
-- check if the URI contains bad words
if ngx.var.uri and
string.match(ngx.var.request_body, "evil")
then
return ngx.redirect("/terms_of_use.html")
end
-- tests passed
}
# proxy_pass/fastcgi_pass/etc settings
}
}
This module embeds LuaJIT 2.0/2.1 into Nginx. It is a core component of OpenResty. If you are using this module, then you are essentially using OpenResty.
Since version v0.10.16 of this module, the standard Lua
interpreter (also known as "PUC-Rio Lua") is not supported anymore. This
document interchangeably uses the terms "Lua" and "LuaJIT" to refer to the
LuaJIT interpreter.
By leveraging Nginx's subrequests, this module allows the integration of the powerful Lua threads (known as Lua "coroutines") into the Nginx event model.
Unlike Apache's mod_lua and Lighttpd's mod_magnet, Lua code executed using this module can be 100% non-blocking on network traffic as long as the Nginx API for Lua provided by this module is used to handle requests to upstream services such as MySQL, PostgreSQL, Memcached, Redis, or upstream HTTP web services.
At least the following Lua libraries and Nginx modules can be used with this module:
Almost any Nginx modules can be used with this ngx_lua module by means of
ngx.location.capture or
ngx.location.capture_multi but it is
recommended to use those lua-resty-* libraries instead of creating
subrequests to access the Nginx upstream modules because the former is usually
much more flexible and memory-efficient.
The Lua interpreter (also known as "Lua State" or "LuaJIT VM instance") is shared across all the requests in a single Nginx worker process to minimize memory use. Request contexts are segregated using lightweight Lua coroutines.
Loaded Lua modules persist in the Nginx worker process level resulting in a small memory footprint in Lua even when under heavy loads.
This module is plugged into Nginx's "http" subsystem so it can only speak downstream communication protocols in the HTTP family (HTTP 0.9/1.0/1.1/2.0, WebSockets, etc...). If you want to do generic TCP communications with the downstream clients, then you should use the ngx_stream_lua module instead, which offers a compatible Lua API.
Just to name a few:
The possibilities are unlimited as the module allows bringing together various elements within Nginx as well as exposing the power of the Lua language to the user. The module provides the full flexibility of scripting while offering performance levels comparable with native C language programs both in terms of CPU time as well as memory footprint thanks to LuaJIT 2.x.
Other scripting language implementations typically struggle to match this performance level.
The latest version of this module is compatible with the following versions of Nginx:
Nginx cores older than 1.6.0 (exclusive) are not supported.
It is highly recommended to use OpenResty releases which bundle Nginx, ngx_lua (this module), LuaJIT, as well as other powerful companion Nginx modules and Lua libraries.
It is discouraged to build this module with Nginx yourself since it is tricky to set up exactly right.
Note that Nginx, LuaJIT, and OpenSSL official releases have various limitations and long-standing bugs that can cause some of this module's features to be disabled, not work properly, or run slower. Official OpenResty releases are recommended because they bundle OpenResty's optimized LuaJIT 2.1 fork and Nginx/OpenSSL patches.
Alternatively, ngx_lua can be manually compiled into Nginx:
Build the source with this module:
wget 'https://openresty.org/download/nginx-1.19.3.tar.gz'
tar -xzvf nginx-1.19.3.tar.gz
cd nginx-1.19.3/
# tell nginx's build system where to find LuaJIT 2.0:
export LUAJIT_LIB=/path/to/luajit/lib
export LUAJIT_INC=/path/to/luajit/include/luajit-2.0
# tell nginx's build system where to find LuaJIT 2.1:
export LUAJIT_LIB=/path/to/luajit/lib
export LUAJIT_INC=/path/to/luajit/include/luajit-2.1
# Here we assume Nginx is to be installed under /opt/nginx/.
./configure --prefix=/opt/nginx \
--with-ld-opt="-Wl,-rpath,/path/to/luajit/lib" \
--add-module=/path/to/ngx_devel_kit \
--add-module=/path/to/lua-nginx-module
# Note that you may also want to add `./configure` options which are used in your
# current nginx build.
# You can get usually those options using command nginx -V
# you can change the parallelism number 2 below to fit the number of spare CPU cores in your
# machine.
make -j2
make install
# Note that this version of lug-nginx-module not allow to set `lua_load_resty_core off;` any more.
# So, you have to install `lua-resty-core` and `lua-resty-lrucache` manually as below.
cd lua-resty-core
make install PREFIX=/opt/nginx
cd lua-resty-lrucache
make install PREFIX=/opt/nginx
# add necessary `lua_package_path` directive to `nginx.conf`, in the http context
lua_package_path "/opt/nginx/lib/lua/?.lua;;";
Starting from NGINX 1.9.11, you can also compile this module as a dynamic module, by using the --add-dynamic-module=PATH option instead of --add-module=PATH on the
./configure command line above. And then you can explicitly load the module in your nginx.conf via the load_module
directive, for example,
load_module /path/to/modules/ndk_http_module.so; # assuming NDK is built as a dynamic module too
load_module /path/to/modules/ngx_http_lua_module.so;
While building this module either via OpenResty or with the Nginx core, you can define the following C macros via the C compiler options:
NGX_LUA_USE_ASSERT
When defined, will enable assertions in the ngx_lua C code base. Recommended for debugging or testing builds. It can introduce some (small) runtime overhead when enabled. This macro was first introduced in the v0.9.10 release.NGX_LUA_ABORT_AT_PANIC
When the LuaJIT VM panics, ngx_lua will instruct the current nginx worker process to quit gracefully by default. By specifying this C macro, ngx_lua will abort the current nginx worker process (which usually results in a core dump file) immediately. This option is useful for debugging VM panics. This option was first introduced in the v0.9.8 release.To enable one or more of these macros, just pass extra C compiler options to the ./configure script of either Nginx or OpenResty. For instance,
./configure --with-cc-opt="-DNGX_LUA_USE_ASSERT -DNGX_LUA_ABORT_AT_PANIC"
The openresty-en mailing list is for English speakers.
The openresty mailing list is for Chinese speakers.
The code repository of this project is hosted on GitHub at openresty/lua-nginx-module.
Please submit bug reports, wishlists, or patches by
Watch YouTube video "Measure Execution Time of Lua Code Correctly in OpenResty"
As from the v0.5.0rc32 release, all *_by_lua_file configure directives (such as content_by_lua_file) support loading LuaJIT 2.0/2.1 raw bytecode files directly:
/path/to/luajit/bin/luajit -b /path/to/input_file.lua /path/to/output_file.ljbc
The -bg option can be used to include debug information in the LuaJIT bytecode file:
/path/to/luajit/bin/luajit -bg /path/to/input_file.lua /path/to/output_file.ljbc
Please refer to the official LuaJIT documentation on the -b option for more details:
https://luajit.org/running.html#opt_b
Note that the bytecode files generated by LuaJIT 2.1 is not compatible with LuaJIT 2.0, and vice versa. The support for LuaJIT 2.1 bytecode was first added in ngx_lua v0.9.3.
Attempts to load standard Lua 5.1 bytecode files into ngx_lua instances linked to LuaJIT 2.0/2.1 (or vice versa) will result in an Nginx error message such as the one below:
[error] 13909#0: *1 failed to load Lua inlined code: bad byte-code header in /path/to/test_file.luac
Loading bytecode files via the Lua primitives like require and
dofile should always work as expected.
If you want to access the system environment variable, say, foo, in Lua via the standard Lua API os.getenv, then you should also list this environment variable name in your nginx.conf file via the env directive. For example,
env foo;
The HTTP 1.0 protocol does not support chunked output and requires an explicit Content-Length header when the response body is not empty in order to support the HTTP 1.0 keep-alive.
So when a HTTP 1.0 request is made and the lua_http10_buffering directive is turned on, ngx_lua will buffer the
output of ngx.say and ngx.print calls and also postpone sending response headers until all the response body output is received.
At that time ngx_lua can calculate the total length of the body and construct a proper Content-Length header to return to the HTTP 1.0 client.
If the Content-Length response header is set in the running Lua code, however, this buffering will be disabled even if the lua_http10_buffering directive is turned on.
For large streaming output responses, it is important to disable the lua_http10_buffering directive to minimise memory usage.
Note that common HTTP benchmark tools such as ab and http_load issue HTTP 1.0 requests by default.
To force curl to send HTTP 1.0 requests, use the -0 option.
With LuaJIT 2.x, it is possible to statically link the bytecode of pure Lua modules into the Nginx executable.
You can use the luajit executable to compile .lua Lua
module files to .o object files containing the exported bytecode
data, and then link the .o files directly in your Nginx build.
Below is a trivial example to demonstrate this. Consider that we have the following .lua file named foo.lua:
-- foo.lua
local _M = {}
function _M.go()
print("Hello from foo")
end
return _M
And then we compile this .lua file to foo.o file:
/path/to/luajit/bin/luajit -bg foo.lua foo.o
What matters here is the name of the .lua file, which determines how you use this module later on the Lua land. The file name foo.o does not matter at all except the .o file extension (which tells luajit what output format is used). If you want to strip the Lua debug information from the resulting bytecode, you can just specify the -b option above instead of -bg.
Then when building Nginx or OpenResty, pass the --with-ld-opt="foo.o" option to the ./configure script:
./configure --with-ld-opt="/path/to/foo.o" ...
Finally, you can just do the following in any Lua code run by ngx_lua:
local foo = require "foo"
foo.go()
And this piece of code no longer depends on the external foo.lua file any more because it has already been compiled into the nginx executable.
If you want to use dot in the Lua module name when calling require, as in
local foo = require "resty.foo"
then you need to rename the foo.lua file to resty_foo.lua before compiling it down to a .o file with the luajit command-line utility.
It is important to use exactly the same version of LuaJIT when compiling .lua files to .o files as building nginx + ngx_lua. This is because the LuaJIT bytecode format may be incompatible between different LuaJIT versions. When the bytecode format is incompatible, you will see a Lua runtime error saying that the Lua module is not found.
When you have multiple .lua files to compile and link, then just specify their .o files at the same time in the value of the --with-ld-opt option. For instance,
./configure --with-ld-opt="/path/to/foo.o /path/to/bar.o" ...
If you have too many .o files, then it might not be feasible to name them all in a single command. In this case, you can build a static library (or archive) for your .o files, as in
ar rcus libmyluafiles.a *.o
then you can link the myluafiles archive as a whole to your nginx executable:
./configure \
--with-ld-opt="-L/path/to/lib -Wl,--whole-archive -lmyluafiles -Wl,--no-whole-archive"
where /path/to/lib is the path of the directory containing the libmyluafiles.a file. It should be noted that the linker option --whole-archive is required here because otherwise our archive will be skipped because no symbols in our archive are mentioned in the main parts of the nginx executable.
To globally share data among all the requests handled by the same Nginx worker
process, encapsulate the shared data into a Lua module, use the Lua
require builtin to import the module, and then manipulate the
shared data in Lua. This works because required Lua modules are loaded only
once and all coroutines will share the same copy of the module (both its code
and data).
Note that the use of global Lua variables is strongly discouraged, as it may lead to unexpected race conditions between concurrent requests.
Here is a small example on sharing data within an Nginx worker via a Lua module:
-- mydata.lua
local _M = {}
local data = {
dog = 3,
cat = 4,
pig = 5,
}
function _M.get_age(name)
return data[name]
end
return _M
and then accessing it from nginx.conf:
location /lua {
content_by_lua_block {
local mydata = require "mydata"
ngx.say(mydata.get_age("dog"))
}
}
The mydata module in this example will only be loaded and run on the first request to the location /lua,
and all subsequent requests to the same Nginx worker process will use the reloaded instance of the
module as well as the same copy of the data in it, until a HUP signal is sent to the Nginx master process to force a reload.
This data sharing technique is essential for high performance Lua applications based on this module.
Note that this data sharing is on a per-worker basis and not on a per-server basis. That is, when there are multiple Nginx worker processes under an Nginx master, data sharing cannot cross the process boundary between these workers.
It is usually recommended to share read-only data this way. You can also share changeable data among all the concurrent requests of each Nginx worker process as long as there is no nonblocking I/O operations (including ngx.sleep) in the middle of your calculations. As long as you do not give the control back to the Nginx event loop and ngx_lua's light thread scheduler (even implicitly), there can never be any race conditions in between. For this reason, always be very careful when you want to share changeable data on the worker level. Buggy optimizations can easily lead to hard-to-debug race conditions under load.
If server-wide data sharing is required, then use one or more of the following approaches:
memcached, redis, MySQL or PostgreSQL. The OpenResty official releases come with a set of companion Nginx modules and Lua libraries that provide interfaces with these data storage mechanisms.The tcpsock:connect method may indicate success despite connection failures such as with Connection Refused errors.
However, later attempts to manipulate the cosocket object will fail and return the actual error status message generated by the failed connect operation.
This issue is due to limitations in the Nginx event model and only appears to affect Mac OS X.
dofile and require builtins are currently implemented as C functions in LuaJIT 2.0/2.1, if the Lua file being loaded by dofile or require invokes ngx.location.capture*, ngx.exec, ngx.exit, or other API functions requiring yielding in the top-level scope of the Lua file, then the Lua error "attempt to yield across C-call boundary" will be raised. To avoid this, put these calls requiring yielding into your own Lua functions in the Lua file instead of the top-level scope of the file.Care must be taken when importing modules, and this form should be used:
local xxx = require('xxx')
instead of the old deprecated form:
require('xxx')
Here is the reason: by design, the global environment has exactly the same lifetime as the Nginx request handler associated with it. Each request handler has its own set of Lua global variables and that is the idea of request isolation. The Lua module is actually loaded by the first Nginx request handler and is cached by the require() built-in in the package.loaded table for later reference, and the module() builtin used by some Lua modules has the side effect of setting a global variable to the loaded module table. But this global variable will be cleared at the end of the request handler, and every subsequent request handler all has its own (clean) global environment. So one will get Lua exception for accessing the nil value.
The use of Lua global variables is a generally inadvisable in the ngx_lua context as:
It is therefore highly recommended to always declare such within an appropriate local scope instead.
-- Avoid
foo = 123
-- Recommended
local foo = 123
-- Avoid
function foo() return 123 end
-- Recommended
local function foo() return 123 end
To find all instances of Lua global variables in your Lua code, run the lua-releng tool across all .lua source files:
$ lua-releng
Checking use of Lua global variables in file lib/foo/bar.lua ...
1 [1489] SETGLOBAL 7 -1 ; contains
55 [1506] GETGLOBAL 7 -3 ; setvar
3 [1545] GETGLOBAL 3 -4 ; varexpand
The output says that the line 1489 of file lib/foo/bar.lua writes to a global variable named contains, the line 1506 reads from the global variable setvar, and line 1545 reads the global varexpand.
This tool will guarantee that local variables in the Lua module functions are all declared with the local keyword, otherwise a runtime exception will be thrown. It prevents undesirable race conditions while accessing such variables. See Data Sharing within an Nginx Worker for the reasons behind this.
The ngx.location.capture and ngx.location.capture_multi directives cannot capture locations that include the add_before_body, add_after_body, auth_request, echo_location, echo_location_async, echo_subrequest, or echo_subrequest_async directives.
location /foo {
content_by_lua_block {
res = ngx.location.capture("/bar")
}
}
location /bar {
echo_location /blah;
}
location /blah {
echo "Success!";
}
$ curl -i http://example.com/foo
will not work as expected.
Due to internal limitations in the Nginx core, the cosocket API is disabled in the following contexts: set_by_lua*, log_by_lua*, header_filter_by_lua*, and body_filter_by_lua.
The cosockets are currently also disabled in the init_by_lua* and init_worker_by_lua* directive contexts but we may add support for these contexts in the future because there is no limitation in the Nginx core (or the limitation might be worked around).
There exists a workaround, however, when the original context does not need to wait for the cosocket results. That is, creating a zero-delay timer via the ngx.timer.at API and do the cosocket results in the timer handler, which runs asynchronously as to the original context creating the timer.
NOTE Following the v0.9.17 release, this pitfall can be avoided by using the *_by_lua_block {} configuration directives.
PCRE sequences such as \d, \s, or \w, require special attention because in string literals, the backslash character, \, is stripped out by both the Lua language parser and by the Nginx config file parser before processing if not within a *_by_lua_block {} directive. So the following snippet will not work as expected:
# nginx.conf
? location /test {
? content_by_lua '
? local regex = "\d+" -- THIS IS WRONG OUTSIDE OF A *_by_lua_block DIRECTIVE
? local m = ngx.re.match("hello, 1234", regex)
? if m then ngx.say(m[0]) else ngx.say("not matched!") end
? ';
? }
# evaluates to "not matched!"
To avoid this, double escape the backslash:
# nginx.conf
location /test {
content_by_lua '
local regex = "\\\\d+"
local m = ngx.re.match("hello, 1234", regex)
if m then ngx.say(m[0]) else ngx.say("not matched!") end
';
}
# evaluates to "1234"
Here, \\\\d+ is stripped down to \\d+ by the Nginx config file parser and this is further stripped down to \d+ by the Lua language parser before running.
Alternatively, the regex pattern can be presented as a long-bracketed Lua string literal by encasing it in "long brackets", [[...]], in which case backslashes have to only be escaped once for the Nginx config file parser.
# nginx.conf
location /test {
content_by_lua '
local regex = [[\\d+]]
local m = ngx.re.match("hello, 1234", regex)
if m then ngx.say(m[0]) else ngx.say("not matched!") end
';
}
# evaluates to "1234"
Here, [[\\d+]] is stripped down to [[\d+]] by the Nginx config file parser and this is processed correctly.
Note that a longer from of the long bracket, [=[...]=], may be required if the regex pattern contains [...] sequences.
The [=[...]=] form may be used as the default form if desired.
# nginx.conf
location /test {
content_by_lua '
local regex = [=[[0-9]+]=]
local m = ngx.re.match("hello, 1234", regex)
if m then ngx.say(m[0]) else ngx.say("not matched!") end
';
}
# evaluates to "1234"
An alternative approach to escaping PCRE sequences is to ensure that Lua code is placed in external script files and executed using the various *_by_lua_file directives.
With this approach, the backslashes are only stripped by the Lua language parser and therefore only need to be escaped once each.
-- test.lua
local regex = "\\d+"
local m = ngx.re.match("hello, 1234", regex)
if m then ngx.say(m[0]) else ngx.say("not matched!") end
-- evaluates to "1234"
Within external script files, PCRE sequences presented as long-bracketed Lua string literals do not require modification.
-- test.lua
local regex = [[\d+]]
local m = ngx.re.match("hello, 1234", regex)
if m then ngx.say(m[0]) else ngx.say("not matched!") end
-- evaluates to "1234"
As noted earlier, PCRE sequences presented within *_by_lua_block {} directives (available following the v0.9.17 release) do not require modification.
# nginx.conf
location /test {
content_by_lua_block {
local regex = [[\d+]]
local m = ngx.re.match("hello, 1234", regex)
if m then ngx.say(m[0]) else ngx.say("not matched!") end
}
}
# evaluates to "1234"
NOTE You are recommended to use by_lua_file when the Lua code is very long.
Mixing SSI with ngx_lua in the same Nginx request is not supported at all. Just use ngx_lua exclusively. Everything you can do with SSI can be done atop ngx_lua anyway and it can be more efficient when using ngx_lua.
Certain Lua APIs provided by ngx_lua do not work in Nginx's SPDY mode yet: ngx.location.capture, ngx.location.capture_multi, and ngx.req.socket.
Nginx may terminate a request early with (at least):
This means that phases that normally run are skipped, such as the rewrite or access phase. This also means that later phases that are run regardless, e.g. log_by_lua, will not have access to information that is normally set in those phases.
bsdrecv method.ngx_hash_t to optimize the built-in header look-up process for ngx.req.set_header, and etc.ignore_resp_headers, ignore_resp_body, and ignore_resp options to ngx.location.capture and ngx.location.capture_multi methods, to allow micro performance tuning on the user side.stat mode similar to mod_lua.The changes made in every release of this module are listed in the change logs of the OpenResty bundle:
https://openresty.org/#Changes
This module uses .travis.yml as the CI configuration.
You can always check .travis.yml for the latest CI configuration.
For developers, you need to run tests locally. You can use util/run-ci.sh
to easily set up the environment and execute the test suite.
To run the Test from the beginning:
git clone https://github.com/openresty/lua-nginx-module.git
cd lua-nginx-module
bash util/run-ci.sh
The following dependencies are required to run the test suite:
Nginx version >= 1.4.2
Perl modules:
Nginx modules:
The order in which these modules are added during configuration is important because the position of any filter module in the filtering chain determines the final output, for example. The correct adding order is shown above.
3rd-party Lua libraries:
Applications:
See also the developer build script for more details on setting up the testing environment.
To run the whole test suite in the default testing mode:
cd /path/to/lua-nginx-module
export PATH=/path/to/your/nginx/sbin:$PATH
prove -I/path/to/test-nginx/lib -r t
To run specific test files:
cd /path/to/lua-nginx-module
export PATH=/path/to/your/nginx/sbin:$PATH
prove -I/path/to/test-nginx/lib t/002-content.t t/003-errors.t
To run a specific test block in a particular test file, add the line --- ONLY to the test block you want to run, and then use the prove utility to run that .t file.
There are also various testing modes based on mockeagain, valgrind, and etc. Refer to the Test::Nginx documentation for more details for various advanced testing modes. See also the test reports for the Nginx test cluster running on Amazon EC2: https://qa.openresty.org.
This module is licensed under the BSD license.
Copyright (C) 2009-2017, by Xiaozhe Wang (chaoslawful) chaoslawful@gmail.com.
Copyright (C) 2009-2025, by Yichun "agentzh" Zhang (章亦春) agentzh@gmail.com, OpenResty Inc.
All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:
Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
Blog posts:
Other related modules and libraries:
The basic building blocks of scripting Nginx with Lua are directives. Directives are used to specify when the user Lua code is run and how the result will be used. Below is a diagram showing the order in which directives are executed.

syntax: lua_load_resty_core on|off
default: lua_load_resty_core on
context: http
This directive is deprecated since the v0.10.16 release of this
module. The resty.core module from
lua-resty-core is now mandatorily
loaded during the Lua VM initialization. Specifying this directive will have no
effect.
This directive was first introduced in the v0.10.15 release and
used to optionally load the resty.core module.
syntax: lua_capture_error_log size
default: none
context: http
Enables a buffer of the specified size for capturing all the Nginx error log message data (not just those produced
by this module or the Nginx http subsystem, but everything) without touching files or disks.
You can use units like k and m in the size value, as in
lua_capture_error_log 100k;
As a rule of thumb, a 4KB buffer can usually hold about 20 typical error log messages. So do the maths!
This buffer never grows. If it is full, new error log messages will replace the oldest ones in the buffer.
The size of the buffer must be bigger than the maximum length of a single error log message (which is 4K in OpenResty and 2K in stock NGINX).
You can read the messages in the buffer on the Lua land via the get_logs() function of the ngx.errlog module of the lua-resty-core library. This Lua API function will return the captured error log messages and also remove these already read from the global capturing buffer, making room for any new error log data. For this reason, the user should not configure this buffer to be too big if the user read the buffered error log data fast enough.
Note that the log level specified in the standard error_log directive does have effect on this capturing facility. It only captures log messages of a level no lower than the specified log level in the error_log directive. The user can still choose to set an even higher filtering log level on the fly via the Lua API function errlog.set_filter_level. So it is more flexible than the static error_log directive.
It is worth noting that there is no way to capture the debugging logs
without building OpenResty or Nginx with the ./configure
option --with-debug. And enabling debugging logs is
strongly discouraged in production builds due to high overhead.
This directive was first introduced in the v0.10.9 release.
syntax: lua_use_default_type on | off
default: lua_use_default_type on
context: http, server, location, location if
Specifies whether to use the MIME type specified by the default_type directive for the default value of the Content-Type response header. Deactivate this directive if a default Content-Type response header for Lua request handlers is not desired.
This directive is turned on by default.
This directive was first introduced in the v0.9.1 release.
syntax: lua_malloc_trim <request-count>
default: lua_malloc_trim 1000
context: http
Asks the underlying libc runtime library to release its cached free memory back to the operating system every
N requests processed by the Nginx core. By default, N is 1000. You can configure the request count
by using your own numbers. Smaller numbers mean more frequent releases, which may introduce higher CPU time consumption and
smaller memory footprint while larger numbers usually lead to less CPU time overhead and relatively larger memory footprint.
Just tune the number for your own use cases.
Configuring the argument to 0 essentially turns off the periodical memory trimming altogether.
lua_malloc_trim 0; # turn off trimming completely
The current implementation uses an Nginx log phase handler to do the request counting. So the appearance of the
log_subrequest on directives in nginx.conf
may make the counting faster when subrequests are involved. By default, only "main requests" count.
Note that this directive does not affect the memory allocated by LuaJIT's own allocator based on the mmap
system call.
This directive was first introduced in the v0.10.7 release.
syntax: lua_code_cache on | off
default: lua_code_cache on
context: http, server, location, location if
Enables or disables the Lua code cache for Lua code in *_by_lua_file directives (like set_by_lua_file and
content_by_lua_file) and Lua modules.
When turning off, every request served by ngx_lua will run in a separate Lua VM instance, starting from the 0.9.3 release. So the Lua files referenced in set_by_lua_file,
content_by_lua_file, access_by_lua_file,
and etc will not be cached
and all Lua modules used will be loaded from scratch. With this in place, developers can adopt an edit-and-refresh approach.
Please note however, that Lua code written inlined within nginx.conf
such as those specified by set_by_lua, content_by_lua,
access_by_lua, and rewrite_by_lua will not be updated when you edit the inlined Lua code in your nginx.conf file because only the Nginx config file parser can correctly parse the nginx.conf
file and the only way is to reload the config file
by sending a HUP signal or just to restart Nginx.
Even when the code cache is enabled, Lua files which are loaded by dofile or loadfile
in *_by_lua_file cannot be cached (unless you cache the results yourself). Usually you can either use the init_by_lua
or init_by_lua_file directives to load all such files or just make these Lua files true Lua modules
and load them via require.
The ngx_lua module does not support the stat mode available with the
Apache mod_lua module (yet).
Disabling the Lua code cache is strongly discouraged for production use and should only be used during development as it has a significant negative impact on overall performance. For example, the performance of a "hello world" Lua example can drop by an order of magnitude after disabling the Lua code cache.
syntax: lua_thread_cache_max_entries <num>
default: lua_thread_cache_max_entries 1024
context: http
Specifies the maximum number of entries allowed in the worker process level lua thread object cache.
This cache recycles the lua thread GC objects among all our "light threads".
A zero value of <num> disables the cache.
Note that this feature requires OpenResty's LuaJIT with the new C API lua_resetthread.
This feature was first introduced in version v0.10.9.
syntax: lua_regex_cache_max_entries <num>
default: lua_regex_cache_max_entries 1024
context: http
Specifies the maximum number of entries allowed in the worker process level compiled regex cache.
The regular expressions used in ngx.re.match, ngx.re.gmatch, ngx.re.sub, and ngx.re.gsub will be cached within this cache if the regex option o (i.e., compile-once flag) is specified.
The default number of entries allowed is 1024 and when this limit is reached, new regular expressions will not be cached (as if the o option was not specified) and there will be one, and only one, warning in the error.log file:
2011/08/27 23:18:26 [warn] 31997#0: *1 lua exceeding regex cache max entries (1024), ...
If you are using the ngx.re.* implementation of lua-resty-core by loading the resty.core.regex module (or just the resty.core module), then an LRU cache is used for the regex cache being used here.
Do not activate the o option for regular expressions (and/or replace string arguments for ngx.re.sub and ngx.re.gsub) that are generated on the fly and give rise to infinite variations to avoid hitting the specified limit.
syntax: lua_regex_match_limit <num>
default: lua_regex_match_limit 0
context: http
Specifies the "match limit" used by the PCRE library when executing the ngx.re API. To quote the PCRE manpage, "the limit ... has the effect of limiting the amount of backtracking that can take place."
When the limit is hit, the error string "pcre_exec() failed: -8" will be returned by the ngx.re API functions on the Lua land.
When setting the limit to 0, the default "match limit" when compiling the PCRE library is used. And this is the default value of this directive.
This directive was first introduced in the v0.8.5 release.
syntax: lua_package_path <lua-style-path-str>
default: The content of LUA_PATH environment variable or Lua's compiled-in defaults.
context: http
Sets the Lua module search path used by scripts specified by set_by_lua,
content_by_lua and others. The path string is in standard Lua path form, and ;;
can be used to stand for the original search paths.
As from the v0.5.0rc29 release, the special notation $prefix or ${prefix} can be used in the search path string to indicate the path of the server prefix usually determined by the -p PATH command-line option while starting the Nginx server.
syntax: lua_package_cpath <lua-style-cpath-str>
default: The content of LUA_CPATH environment variable or Lua's compiled-in defaults.
context: http
Sets the Lua C-module search path used by scripts specified by set_by_lua,
content_by_lua and others. The cpath string is in standard Lua cpath form, and ;;
can be used to stand for the original cpath.
As from the v0.5.0rc29 release, the special notation $prefix or ${prefix} can be used in the search path string to indicate the path of the server prefix usually determined by the -p PATH command-line option while starting the Nginx server.
syntax: init_by_lua <lua-script-str>
context: http
phase: loading-config
NOTE Use of this directive is discouraged following the v0.9.17 release. Use the init_by_lua_block directive instead.
Similar to the init_by_lua_block directive, but accepts the Lua source directly in an Nginx string literal (which requires special character escaping).
For instance,
init_by_lua '
print("I need no extra escaping here, for example: \r\nblah")
'
This directive was first introduced in the v0.5.5 release.
syntax: init_by_lua_block { lua-script }
context: http
phase: loading-config
When Nginx receives the HUP signal and starts reloading the config file, the Lua VM will also be re-created and init_by_lua_block will run again on the new Lua VM. In case that the lua_code_cache directive is turned off (default on), the init_by_lua_block handler will run upon every request because in this special mode a standalone Lua VM is always created for each request.
Usually you can pre-load Lua modules at server start-up by means of this hook and take advantage of modern operating systems' copy-on-write (COW) optimization. Here is an example for pre-loading Lua modules:
# this runs before forking out nginx worker processes:
init_by_lua_block { require "cjson" }
server {
location = /api {
content_by_lua_block {
-- the following require() will just return
-- the already loaded module from package.loaded:
ngx.say(require "cjson".encode{dog = 5, cat = 6})
}
}
}
You can also initialize the lua_shared_dict shm storage at this phase. Here is an example for this:
lua_shared_dict dogs 1m;
init_by_lua_block {
local dogs = ngx.shared.dogs
dogs:set("Tom", 56)
}
server {
location = /api {
content_by_lua_block {
local dogs = ngx.shared.dogs
ngx.say(dogs:get("Tom"))
}
}
}
But note that, the lua_shared_dict's shm storage will not be cleared through a config reload (via the HUP signal, for example). So if you do not want to re-initialize the shm storage in your init_by_lua_block code in this case, then you just need to set a custom flag in the shm storage and always check the flag in your init_by_lua_block code.
Because the Lua code in this context runs before Nginx forks its worker processes (if any), data or code loaded here will enjoy the Copy-on-write (COW) feature provided by many operating systems among all the worker processes, thus saving a lot of memory.
Do not initialize your own Lua global variables in this context because use of Lua global variables have performance penalties and can lead to global namespace pollution (see the Lua Variable Scope section for more details). The recommended way is to use proper Lua module files (but do not use the standard Lua function module() to define Lua modules because it pollutes the global namespace as well) and call require() to load your own module files in init_by_lua_block or other contexts (require() does cache the loaded Lua modules in the global package.loaded table in the Lua registry so your modules will only loaded once for the whole Lua VM instance).
Only a small set of the Nginx API for Lua is supported in this context:
More Nginx APIs for Lua may be supported in this context upon future user requests.
Basically you can safely use Lua libraries that do blocking I/O in this very context because blocking the master process during server start-up is completely okay. Even the Nginx core does blocking I/O (at least on resolving upstream's host names) at the configure-loading phase.
You should be very careful about potential security vulnerabilities in your Lua code registered in this context because the Nginx master process is often run under the root account.
This directive was first introduced in the v0.9.17 release.
See also the following blog posts for more details on OpenResty and Nginx's shared memory zones:
syntax: init_by_lua_file <path-to-lua-script-file>
context: http
phase: loading-config
Equivalent to init_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code or LuaJIT bytecode to be executed.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
This directive was first introduced in the v0.5.5 release.
syntax: init_worker_by_lua <lua-script-str>
context: http
phase: starting-worker
NOTE Use of this directive is discouraged following the v0.9.17 release. Use the init_worker_by_lua_block directive instead.
Similar to the init_worker_by_lua_block directive, but accepts the Lua source directly in an Nginx string literal (which requires special character escaping).
For instance,
init_worker_by_lua '
print("I need no extra escaping here, for example: \r\nblah")
';
This directive was first introduced in the v0.9.5 release.
This hook no longer runs in the cache manager and cache loader processes since the v0.10.12 release.
syntax: init_worker_by_lua_block { lua-script }
context: http
phase: starting-worker
Runs the specified Lua code upon every Nginx worker process's startup when the master process is enabled. When the master process is disabled, this hook will just run after init_by_lua*.
This hook is often used to create per-worker reoccurring timers (via the ngx.timer.at Lua API), either for backend health-check or other timed routine work. Below is an example,
init_worker_by_lua_block {
local delay = 3 -- in seconds
local new_timer = ngx.timer.at
local log = ngx.log
local ERR = ngx.ERR
local check
check = function(premature)
if not premature then
-- do the health check or other routine work
local ok, err = new_timer(delay, check)
if not ok then
log(ERR, "failed to create timer: ", err)
return
end
end
-- do something in timer
end
local hdl, err = new_timer(delay, check)
if not hdl then
log(ERR, "failed to create timer: ", err)
return
end
-- other job in init_worker_by_lua
}
This directive was first introduced in the v0.9.17 release.
This hook no longer runs in the cache manager and cache loader processes since the v0.10.12 release.
syntax: init_worker_by_lua_file <lua-file-path>
context: http
phase: starting-worker
Similar to init_worker_by_lua_block, but accepts the file path to a Lua source file or Lua bytecode file.
This directive was first introduced in the v0.9.5 release.
This hook no longer runs in the cache manager and cache loader processes since the v0.10.12 release.
syntax: exit_worker_by_lua_block { lua-script }
context: http
phase: exiting-worker
Runs the specified Lua code upon every Nginx worker process's exit when the master process is enabled. When the master process is disabled, this hook will run before the Nginx process exits.
This hook is often used to release resources allocated by each worker (e.g. resources allocated by init_worker_by_lua*), or to prevent workers from exiting abnormally.
For example,
exit_worker_by_lua_block {
print("log from exit_worker_by_lua_block")
}
It's not allowed to create a timer (even a 0-delay timer) here since it runs after all timers have been processed.
This directive was first introduced in the v0.10.18 release.
syntax: exit_worker_by_lua_file <path-to-lua-script-file>
context: http
phase: exiting-worker
Similar to exit_worker_by_lua_block, but accepts the file path to a Lua source file or Lua bytecode file.
This directive was first introduced in the v0.10.18 release.
syntax: set_by_lua $res <lua-script-str> [$arg1 $arg2 ...]
context: server, server if, location, location if
phase: rewrite
NOTE Use of this directive is discouraged following the v0.9.17 release. Use the set_by_lua_block directive instead.
Similar to the set_by_lua_block directive, but accepts the Lua source directly in an Nginx string literal (which requires special character escaping), and
For example,
set_by_lua $res ' return 32 + math.cos(32) ';
# $res now has the value "32.834223360507" or alike.
As from the v0.5.0rc29 release, Nginx variable interpolation is disabled in the <lua-script-str> argument of this directive and therefore, the dollar sign character ($) can be used directly.
This directive requires the ngx_devel_kit module.
syntax: set_by_lua_block $res { lua-script }
context: server, server if, location, location if
phase: rewrite
Executes code specified inside a pair of curly braces ({}), and returns string output to $res.
The code inside a pair of curly braces ({}) can make API calls and can retrieve input arguments from the ngx.arg table (index starts from 1 and increases sequentially).
This directive is designed to execute short, fast running code blocks as the Nginx event loop is blocked during code execution. Time consuming code sequences should therefore be avoided.
This directive is implemented by injecting custom commands into the standard ngx_http_rewrite_module's command list. Because ngx_http_rewrite_module does not support nonblocking I/O in its commands, Lua APIs requiring yielding the current Lua "light thread" cannot work in this directive.
At least the following API functions are currently disabled within the context of set_by_lua_block:
In addition, note that this directive can only write out a value to a single Nginx variable at a time. However, a workaround is possible using the ngx.var.VARIABLE interface.
location /foo {
set $diff ''; # we have to predefine the $diff variable here
set_by_lua_block $sum {
local a = 32
local b = 56
ngx.var.diff = a - b -- write to $diff directly
return a + b -- return the $sum value normally
}
echo "sum = $sum, diff = $diff";
}
This directive can be freely mixed with all directives of the ngx_http_rewrite_module, set-misc-nginx-module, and array-var-nginx-module modules. All of these directives will run in the same order as they appear in the config file.
set $foo 32;
set_by_lua_block $bar { return tonumber(ngx.var.foo) + 1 }
set $baz "bar: $bar"; # $baz == "bar: 33"
No special escaping is required in the Lua code block.
This directive requires the ngx_devel_kit module.
This directive was first introduced in the v0.9.17 release.
syntax: set_by_lua_file $res <path-to-lua-script-file> [$arg1 $arg2 ...]
context: server, server if, location, location if
phase: rewrite
Equivalent to set_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or, as from the v0.5.0rc32 release, the LuaJIT bytecode to be executed.
Nginx variable interpolation is supported in the <path-to-lua-script-file> argument string of this directive. But special care must be taken for injection attacks.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
When the Lua code cache is turned on (by default), the user code is loaded once at the first request and cached
and the Nginx config must be reloaded each time the Lua source file is modified.
The Lua code cache can be temporarily disabled during development by
switching lua_code_cache off in nginx.conf to avoid reloading Nginx.
This directive requires the ngx_devel_kit module.
syntax: precontent_by_lua_block { lua-script }
context: http, server, location, location if
phase: precontent tail
Acts as a precontent phase handler and executes Lua code string specified in { <lua-script } for every request.
The Lua code may make API calls and is executed as a new spawned coroutine in an independent global environment (i.e. a sandbox).
Note that this handler always runs after the standard ngx_http_mirror_module and ngx_http_try_files_module. For example:
location /images/ {
try_files $uri /images/default.gif;
precontent_by_lua_block {
ngx.log(ngx.NOTICE, "file found")
}
}
location = /images/default.gif {
expires 30s;
precontent_by_lua_block {
ngx.log(ngx.NOTICE, "file not found, use default.gif instead")
}
}
That is, if a request for /images/foo.jpg comes in and the file does not exist, the request will be internally redirected to /images/default.gif before precontent_by_lua_block, and then the precontent_by_lua_block in new location will run and log "file not found, use default.gif instead".
You can use precontent_by_lua_block to perform some preparatory functions after the access phase handler but before the proxy or other content handler. Especially some functions that cannot be performed in balancer_by_lua_block.
you can use the precontent_by_lua_no_postpone directive to control when to run this handler inside the "precontent" request-processing phase of Nginx.
syntax: precontent_by_lua_file <path-to-lua-script-file>
context: http, server, location, location if
phase: precontent tail
Equivalent to precontent_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or, as from the v0.5.0rc32 release, the LuaJIT bytecode to be executed.
Nginx variables can be used in the <path-to-lua-script-file> string to provide flexibility. This however carries some risks and is not ordinarily recommended.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
When the Lua code cache is turned on (by default), the user code is loaded once at the first request and cached
and the Nginx config must be reloaded each time the Lua source file is modified.
The Lua code cache can be temporarily disabled during development by switching lua_code_cache off in nginx.conf to avoid repeatedly reloading Nginx.
Nginx variables are supported in the file path for dynamic dispatch just as in content_by_lua_file.
But be very careful about malicious user inputs and always carefully validate or filter out the user-supplied path components.
syntax: content_by_lua <lua-script-str>
context: location, location if
phase: content
NOTE Use of this directive is discouraged following the v0.9.17 release. Use the content_by_lua_block directive instead.
Similar to the content_by_lua_block directive, but accepts the Lua source directly in an Nginx string literal (which requires special character escaping).
For instance,
content_by_lua '
ngx.say("I need no extra escaping here, for example: \r\nblah")
';
syntax: content_by_lua_block { lua-script }
context: location, location if
phase: content
For instance,
content_by_lua_block {
ngx.say("I need no extra escaping here, for example: \r\nblah")
}
Acts as a "content handler" and executes Lua code string specified in { lua-script } for every request.
The Lua code may make API calls and is executed as a new spawned coroutine in an independent global environment (i.e. a sandbox).
Do not use this directive and other content handler directives in the same location. For example, this directive and the proxy_pass directive should not be used in the same location.
This directive was first introduced in the v0.9.17 release.
syntax: content_by_lua_file <path-to-lua-script-file>
context: location, location if
phase: content
Equivalent to content_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or, as from the v0.5.0rc32 release, the LuaJIT bytecode to be executed.
If the file is not found, a 404 Not Found status code will be returned, and a 503 Service Temporarily Unavailable status code will be returned in case of errors in reading other files.
Nginx variables can be used in the <path-to-lua-script-file> string to provide flexibility. This however carries some risks and is not ordinarily recommended.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
When the Lua code cache is turned on (by default), the user code is loaded once at the first request and cached
and the Nginx config must be reloaded each time the Lua source file is modified.
The Lua code cache can be temporarily disabled during development by
switching lua_code_cache off in nginx.conf to avoid reloading Nginx.
Nginx variables are supported in the file path for dynamic dispatch, for example:
# CAUTION: contents in nginx var must be carefully filtered,
# otherwise there'll be great security risk!
location ~ ^/app/([-_a-zA-Z0-9/]+) {
set $path $1;
content_by_lua_file /path/to/lua/app/root/$path.lua;
}
But be very careful about malicious user inputs and always carefully validate or filter out the user-supplied path components.
syntax: server_rewrite_by_lua_block { lua-script }
context: http, server
phase: server rewrite
Acts as a server rewrite phase handler and executes Lua code string specified in { lua-script } for every request.
The Lua code may make API calls and is executed as a new spawned coroutine in an independent global environment (i.e. a sandbox).
server {
...
server_rewrite_by_lua_block {
ngx.ctx.a = "server_rewrite_by_lua_block in http"
}
location /lua {
content_by_lua_block {
ngx.say(ngx.ctx.a)
ngx.log(ngx.INFO, ngx.ctx.a)
}
}
}
Just as any other rewrite phase handlers, server_rewrite_by_lua_block also runs in subrequests.
server {
server_rewrite_by_lua_block {
ngx.log(ngx.INFO, "is_subrequest:", ngx.is_subrequest)
}
location /lua {
content_by_lua_block {
local res = ngx.location.capture("/sub")
ngx.print(res.body)
}
}
location /sub {
content_by_lua_block {
ngx.say("OK")
}
}
}
Note that when calling ngx.exit(ngx.OK) within a server_rewrite_by_lua_block handler, the Nginx request processing control flow will still continue to the content handler. To terminate the current request from within a server_rewrite_by_lua_block handler, call ngx.exit with status >= 200 (ngx.HTTP_OK) and status < 300 (ngx.HTTP_SPECIAL_RESPONSE) for successful quits and ngx.exit(ngx.HTTP_INTERNAL_SERVER_ERROR) (or its friends) for failures.
server_rewrite_by_lua_block {
ngx.exit(503)
}
location /bar {
...
# never exec
}
syntax: server_rewrite_by_lua_file <path-to-lua-script-file>
context: http, server
phase: server rewrite
Equivalent to server_rewrite_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or, as from the v0.10.22 release, the LuaJIT bytecode to be executed.
Nginx variables can be used in the <path-to-lua-script-file> string to provide flexibility. This however carries some risks and is not ordinarily recommended.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
When the Lua code cache is turned on (by default), the user code is loaded once at the first request and cached and the Nginx config must be reloaded each time the Lua source file is modified. The Lua code cache can be temporarily disabled during development by switching lua_code_cache off in nginx.conf to avoid reloading Nginx.
syntax: rewrite_by_lua <lua-script-str>
context: http, server, location, location if
phase: rewrite tail
NOTE Use of this directive is discouraged following the v0.9.17 release. Use the rewrite_by_lua_block directive instead.
Similar to the rewrite_by_lua_block directive, but accepts the Lua source directly in an Nginx string literal (which requires special character escaping).
For instance,
rewrite_by_lua '
do_something("hello, world!\nhiya\n")
';
syntax: rewrite_by_lua_block { lua-script }
context: http, server, location, location if
phase: rewrite tail
Acts as a rewrite phase handler and executes Lua code string specified in { lua-script } for every request.
The Lua code may make API calls and is executed as a new spawned coroutine in an independent global environment (i.e. a sandbox).
Note that this handler always runs after the standard ngx_http_rewrite_module. So the following will work as expected:
location /foo {
set $a 12; # create and initialize $a
set $b ""; # create and initialize $b
rewrite_by_lua_block {
ngx.var.b = tonumber(ngx.var.a) + 1
}
echo "res = $b";
}
because set $a 12 and set $b "" run before rewrite_by_lua_block.
On the other hand, the following will not work as expected:
? location /foo {
? set $a 12; # create and initialize $a
? set $b ''; # create and initialize $b
? rewrite_by_lua_block {
? ngx.var.b = tonumber(ngx.var.a) + 1
? }
? if ($b = '13') {
? rewrite ^ /bar redirect;
? break;
? }
?
? echo "res = $b";
? }
because if runs before rewrite_by_lua_block even if it is placed after rewrite_by_lua_block in the config.
The right way of doing this is as follows:
location /foo {
set $a 12; # create and initialize $a
set $b ''; # create and initialize $b
rewrite_by_lua_block {
ngx.var.b = tonumber(ngx.var.a) + 1
if tonumber(ngx.var.b) == 13 then
return ngx.redirect("/bar")
end
}
echo "res = $b";
}
Note that the ngx_eval module can be approximated by using rewrite_by_lua_block. For example,
location / {
eval $res {
proxy_pass http://foo.com/check-spam;
}
if ($res = 'spam') {
rewrite ^ /terms-of-use.html redirect;
}
fastcgi_pass ...;
}
can be implemented in ngx_lua as:
location = /check-spam {
internal;
proxy_pass http://foo.com/check-spam;
}
location / {
rewrite_by_lua_block {
local res = ngx.location.capture("/check-spam")
if res.body == "spam" then
return ngx.redirect("/terms-of-use.html")
end
}
fastcgi_pass ...;
}
Just as any other rewrite phase handlers, rewrite_by_lua_block also runs in subrequests.
Note that when calling ngx.exit(ngx.OK) within a rewrite_by_lua_block handler, the Nginx request processing control flow will still continue to the content handler. To terminate the current request from within a rewrite_by_lua_block handler, call ngx.exit with status >= 200 (ngx.HTTP_OK) and status < 300 (ngx.HTTP_SPECIAL_RESPONSE) for successful quits and ngx.exit(ngx.HTTP_INTERNAL_SERVER_ERROR) (or its friends) for failures.
If the ngx_http_rewrite_module's rewrite directive is used to change the URI and initiate location re-lookups (internal redirections), then any rewrite_by_lua_block or rewrite_by_lua_file_block code sequences within the current location will not be executed. For example,
location /foo {
rewrite ^ /bar;
rewrite_by_lua_block {
ngx.exit(503)
}
}
location /bar {
...
}
Here the Lua code ngx.exit(503) will never run. This will be the case if rewrite ^ /bar last is used as this will similarly initiate an internal redirection. If the break modifier is used instead, there will be no internal redirection and the rewrite_by_lua_block code will be executed.
The rewrite_by_lua_block code will always run at the end of the rewrite request-processing phase unless rewrite_by_lua_no_postpone is turned on.
This directive was first introduced in the v0.9.17 release.
syntax: rewrite_by_lua_file <path-to-lua-script-file>
context: http, server, location, location if
phase: rewrite tail
Equivalent to rewrite_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or, as from the v0.5.0rc32 release, the LuaJIT bytecode to be executed.
Nginx variables can be used in the <path-to-lua-script-file> string to provide flexibility. This however carries some risks and is not ordinarily recommended.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
When the Lua code cache is turned on (by default), the user code is loaded once at the first request and cached and the Nginx config must be reloaded each time the Lua source file is modified. The Lua code cache can be temporarily disabled during development by switching lua_code_cache off in nginx.conf to avoid reloading Nginx.
The rewrite_by_lua_file code will always run at the end of the rewrite request-processing phase unless rewrite_by_lua_no_postpone is turned on.
Nginx variables are supported in the file path for dynamic dispatch just as in content_by_lua_file.
syntax: access_by_lua <lua-script-str>
context: http, server, location, location if
phase: access tail
NOTE Use of this directive is discouraged following the v0.9.17 release. Use the access_by_lua_block directive instead.
Similar to the access_by_lua_block directive, but accepts the Lua source directly in an Nginx string literal (which requires special character escaping).
For instance,
access_by_lua '
do_something("hello, world!\nhiya\n")
';
syntax: access_by_lua_block { lua-script }
context: http, server, location, location if
phase: access tail
Acts as an access phase handler and executes Lua code string specified in { <lua-script } for every request.
The Lua code may make API calls and is executed as a new spawned coroutine in an independent global environment (i.e. a sandbox).
Note that this handler always runs after the standard ngx_http_access_module. So the following will work as expected:
location / {
deny 192.168.1.1;
allow 192.168.1.0/24;
allow 10.1.1.0/16;
deny all;
access_by_lua_block {
local res = ngx.location.capture("/mysql", { ... })
...
}
# proxy_pass/fastcgi_pass/...
}
That is, if a client IP address is in the blacklist, it will be denied before the MySQL query for more complex authentication is executed by access_by_lua_block.
Note that the ngx_auth_request module can be approximated by using access_by_lua_block:
location / {
auth_request /auth;
# proxy_pass/fastcgi_pass/postgres_pass/...
}
can be implemented in ngx_lua as:
location / {
access_by_lua_block {
local res = ngx.location.capture("/auth")
if res.status == ngx.HTTP_OK then
return
end
if res.status == ngx.HTTP_FORBIDDEN then
ngx.exit(res.status)
end
ngx.exit(ngx.HTTP_INTERNAL_SERVER_ERROR)
}
# proxy_pass/fastcgi_pass/postgres_pass/...
}
As with other access phase handlers, access_by_lua_block will not run in subrequests.
Note that when calling ngx.exit(ngx.OK) within a access_by_lua_block handler, the Nginx request processing control flow will still continue to the content handler. To terminate the current request from within a access_by_lua_block handler, call ngx.exit with status >= 200 (ngx.HTTP_OK) and status < 300 (ngx.HTTP_SPECIAL_RESPONSE) for successful quits and ngx.exit(ngx.HTTP_INTERNAL_SERVER_ERROR) (or its friends) for failures.
Starting from the v0.9.20 release, you can use the access_by_lua_no_postpone
directive to control when to run this handler inside the "access" request-processing phase
of Nginx.
This directive was first introduced in the v0.9.17 release.
syntax: access_by_lua_file <path-to-lua-script-file>
context: http, server, location, location if
phase: access tail
Equivalent to access_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or, as from the v0.5.0rc32 release, the LuaJIT bytecode to be executed.
Nginx variables can be used in the <path-to-lua-script-file> string to provide flexibility. This however carries some risks and is not ordinarily recommended.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
When the Lua code cache is turned on (by default), the user code is loaded once at the first request and cached
and the Nginx config must be reloaded each time the Lua source file is modified.
The Lua code cache can be temporarily disabled during development by switching lua_code_cache off in nginx.conf to avoid repeatedly reloading Nginx.
Nginx variables are supported in the file path for dynamic dispatch just as in content_by_lua_file.
syntax: header_filter_by_lua <lua-script-str>
context: http, server, location, location if
phase: output-header-filter
NOTE Use of this directive is discouraged following the v0.9.17 release. Use the header_filter_by_lua_block directive instead.
Similar to the header_filter_by_lua_block directive, but accepts the Lua source directly in an Nginx string literal (which requires special character escaping).
For instance,
header_filter_by_lua '
ngx.header["content-length"] = nil
';
This directive was first introduced in the v0.2.1rc20 release.
syntax: header_filter_by_lua_block { lua-script }
context: http, server, location, location if
phase: output-header-filter
Uses Lua code specified in { lua-script } to define an output header filter.
Note that the following API functions are currently disabled within this context:
Here is an example of overriding a response header (or adding one if absent) in our Lua header filter:
location / {
proxy_pass http://mybackend;
header_filter_by_lua_block {
ngx.header.Foo = "blah"
}
}
This directive was first introduced in the v0.9.17 release.
syntax: header_filter_by_lua_file <path-to-lua-script-file>
context: http, server, location, location if
phase: output-header-filter
Equivalent to header_filter_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or as from the v0.5.0rc32 release, the LuaJIT bytecode to be executed.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
This directive was first introduced in the v0.2.1rc20 release.
syntax: body_filter_by_lua <lua-script-str>
context: http, server, location, location if
phase: output-body-filter
NOTE Use of this directive is discouraged following the v0.9.17 release. Use the body_filter_by_lua_block directive instead.
Similar to the body_filter_by_lua_block directive, but accepts the Lua source directly in an Nginx string literal (which requires special character escaping).
For instance,
body_filter_by_lua '
local data, eof = ngx.arg[1], ngx.arg[2]
';
This directive was first introduced in the v0.5.0rc32 release.
syntax: body_filter_by_lua_block { lua-script-str }
context: http, server, location, location if
phase: output-body-filter
Uses Lua code specified in { lua-script } to define an output body filter.
The input data chunk is passed via ngx.arg[1] (as a Lua string value) and the "eof" flag indicating the end of the response body data stream is passed via ngx.arg[2] (as a Lua boolean value).
Behind the scene, the "eof" flag is just the last_buf (for main requests) or last_in_chain (for subrequests) flag of the Nginx chain link buffers. (Before the v0.7.14 release, the "eof" flag does not work at all in subrequests.)
The output data stream can be aborted immediately by running the following Lua statement:
return ngx.ERROR
This will truncate the response body and usually result in incomplete and also invalid responses.
The Lua code can pass its own modified version of the input data chunk to the downstream Nginx output body filters by overriding ngx.arg[1] with a Lua string or a Lua table of strings. For example, to transform all the lowercase letters in the response body, we can just write:
location / {
proxy_pass http://mybackend;
body_filter_by_lua_block {
ngx.arg[1] = string.upper(ngx.arg[1])
}
}
When setting nil or an empty Lua string value to ngx.arg[1], no data chunk will be passed to the downstream Nginx output filters at all.
Likewise, new "eof" flag can also be specified by setting a boolean value to ngx.arg[2]. For example,
location /t {
echo hello world;
echo hiya globe;
body_filter_by_lua_block {
local chunk = ngx.arg[1]
if string.match(chunk, "hello") then
ngx.arg[2] = true -- new eof
return
end
-- just throw away any remaining chunk data
ngx.arg[1] = nil
}
}
Then GET /t will just return the output
hello world
That is, when the body filter sees a chunk containing the word "hello", then it will set the "eof" flag to true immediately, resulting in truncated but still valid responses.
When the Lua code may change the length of the response body, then it is required to always clear out the Content-Length response header (if any) in a header filter to enforce streaming output, as in
location /foo {
# fastcgi_pass/proxy_pass/...
header_filter_by_lua_block {
ngx.header.content_length = nil
}
body_filter_by_lua_block {
ngx.arg[1] = string.len(ngx.arg[1]) .. "\n"
}
}
Note that the following API functions are currently disabled within this context due to the limitations in Nginx output filter's current implementation:
Nginx output filters may be called multiple times for a single request because response body may be delivered in chunks. Thus, the Lua code specified by in this directive may also run multiple times in the lifetime of a single HTTP request.
This directive was first introduced in the v0.9.17 release.
syntax: body_filter_by_lua_file <path-to-lua-script-file>
context: http, server, location, location if
phase: output-body-filter
Equivalent to body_filter_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or, as from the v0.5.0rc32 release, the LuaJIT bytecode to be executed.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
This directive was first introduced in the v0.5.0rc32 release.
syntax: log_by_lua <lua-script-str>
context: http, server, location, location if
phase: log
NOTE Use of this directive is discouraged following the v0.9.17 release. Use the log_by_lua_block directive instead.
Similar to the log_by_lua_block directive, but accepts the Lua source directly in an Nginx string literal (which requires special character escaping).
For instance,
log_by_lua '
print("I need no extra escaping here, for example: \r\nblah")
';
This directive was first introduced in the v0.5.0rc31 release.
syntax: log_by_lua_block { lua-script }
context: http, server, location, location if
phase: log
Runs the Lua source code inlined as the { lua-script } at the log request processing phase. This does not replace the current access logs, but runs before.
Note that the following API functions are currently disabled within this context:
Here is an example of gathering average data for $upstream_response_time:
lua_shared_dict log_dict 5M;
server {
location / {
proxy_pass http://mybackend;
log_by_lua_block {
local log_dict = ngx.shared.log_dict
local upstream_time = tonumber(ngx.var.upstream_response_time)
local sum = log_dict:get("upstream_time-sum") or 0
sum = sum + upstream_time
log_dict:set("upstream_time-sum", sum)
local newval, err = log_dict:incr("upstream_time-nb", 1)
if not newval and err == "not found" then
log_dict:add("upstream_time-nb", 0)
log_dict:incr("upstream_time-nb", 1)
end
}
}
location = /status {
content_by_lua_block {
local log_dict = ngx.shared.log_dict
local sum = log_dict:get("upstream_time-sum")
local nb = log_dict:get("upstream_time-nb")
if nb and sum then
ngx.say("average upstream response time: ", sum / nb,
" (", nb, " reqs)")
else
ngx.say("no data yet")
end
}
}
}
This directive was first introduced in the v0.9.17 release.
syntax: log_by_lua_file <path-to-lua-script-file>
context: http, server, location, location if
phase: log
Equivalent to log_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or, as from the v0.5.0rc32 release, the LuaJIT bytecode to be executed.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
This directive was first introduced in the v0.5.0rc31 release.
syntax: balancer_by_lua_block { lua-script }
context: upstream
phase: content
This directive runs Lua code as an upstream balancer for any upstream entities defined
by the upstream {} configuration block.
For instance,
upstream foo {
server 127.0.0.1;
balancer_by_lua_block {
-- use Lua to do something interesting here
-- as a dynamic balancer
}
}
server {
location / {
proxy_pass http://foo;
}
}
The resulting Lua load balancer can work with any existing Nginx upstream modules like ngx_proxy and ngx_fastcgi.
Also, the Lua load balancer can work with the standard upstream connection pool mechanism,
i.e., the standard keepalive directive.
Just ensure that the keepalive directive
is used after this balancer_by_lua_block directive in a single upstream {} configuration block.
The Lua load balancer can totally ignore the list of servers defined in the upstream {} block
and select peer from a completely dynamic server list (even changing per request) via the
ngx.balancer module
from the lua-resty-core library.
The Lua code handler registered by this directive might get called more than once in a single downstream request when the Nginx upstream mechanism retries the request on conditions specified by directives like the proxy_next_upstream directive.
This Lua code execution context does not support yielding, so Lua APIs that may yield (like cosockets and "light threads") are disabled in this context. One can usually work around this limitation by doing such operations in an earlier phase handler (like precontent_by_lua*) and passing along the result into this context via the ngx.ctx table.
This directive was first introduced in the v0.10.0 release.
syntax: balancer_by_lua_file <path-to-lua-script-file>
context: upstream
phase: content
Equivalent to balancer_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or, as from the v0.5.0rc32 release, the LuaJIT bytecode to be executed.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
This directive was first introduced in the v0.10.0 release.
syntax: balancer_keepalive <total-connections>
context: upstream
phase: loading-config
The total-connections parameter sets the maximum number of idle
keepalive connections to upstream servers that are preserved in the cache of
each worker process. When this number is exceeded, the least recently used
connections are closed.
It should be particularly noted that the keepalive directive does not limit the total number of connections to upstream servers that an nginx worker process can open. The connections parameter should be set to a number small enough to let upstream servers process new incoming connections as well.
This directive was first introduced in the v0.10.21 release.
syntax: lua_need_request_body <on|off>
default: off
context: http, server, location, location if
phase: depends on usage
Determines whether to force the request body data to be read before running rewrite/access/content_by_lua* or not. The Nginx core does not read the client request body by default and if request body data is required, then this directive should be turned on or the ngx.req.read_body function should be called within the Lua code.
To read the request body data within the $request_body variable, client_body_buffer_size must have the same value as client_max_body_size. Because when the content length exceeds client_body_buffer_size but less than client_max_body_size, Nginx will buffer the data into a temporary file on the disk, which will lead to empty value in the $request_body variable.
If the current location includes rewrite_by_lua* directives,
then the request body will be read just before the rewrite_by_lua* code is run (and also at the
rewrite phase). Similarly, if only content_by_lua is specified,
the request body will not be read until the content handler's Lua code is
about to run (i.e., the request body will be read during the content phase).
It is recommended however, to use the ngx.req.read_body and ngx.req.discard_body functions for finer control over the request body reading process instead.
This also applies to access_by_lua*.
syntax: ssl_client_hello_by_lua_block { lua-script }
context: http, server
phase: right-after-client-hello-message-was-processed
This directive runs user Lua code when Nginx is about to post-process the SSL client hello message for the downstream SSL (https) connections.
It is particularly useful for dynamically setting the SSL protocols according to the SNI.
It is also useful to do some custom operations according to the per-connection information in the client hello message.
For example, one can parse custom client hello extension and do the corresponding handling in pure Lua.
This Lua handler will always run whether the SSL session is resumed (via SSL session IDs or TLS session tickets) or not.
While the ssl_certificate_by_lua* Lua handler will only runs when initiating a full SSL handshake.
The ngx.ssl.clienthello Lua modules provided by the lua-resty-core library are particularly useful in this context.
Note that this handler runs in extremely early stage of SSL handshake, before the SSL client hello extensions are parsed.
So you can not use some Lua API like ssl.server_name() which is dependent on the later stage's processing.
Also note that only the directive in default server is valid for several virtual servers with the same IP address and port.
Below is a trivial example using the ngx.ssl.clienthello module at the same time:
server {
listen 443 ssl;
server_name test.com;
ssl_certificate /path/to/cert.crt;
ssl_certificate_key /path/to/key.key;
ssl_client_hello_by_lua_block {
local ssl_clt = require "ngx.ssl.clienthello"
local host, err = ssl_clt.get_client_hello_server_name()
if host == "test.com" then
ssl_clt.set_protocols({"TLSv1", "TLSv1.1"})
elseif host == "test2.com" then
ssl_clt.set_protocols({"TLSv1.2", "TLSv1.3"})
elseif not host then
ngx.log(ngx.ERR, "failed to get the SNI name: ", err)
ngx.exit(ngx.ERROR)
else
ngx.log(ngx.ERR, "unknown SNI name: ", host)
ngx.exit(ngx.ERROR)
end
}
...
}
server {
listen 443 ssl;
server_name test2.com;
ssl_certificate /path/to/cert.crt;
ssl_certificate_key /path/to/key.key;
...
}
See more information in the ngx.ssl.clienthello Lua modules' official documentation.
Uncaught Lua exceptions in the user Lua code immediately abort the current SSL session, so does the
ngx.exit call with an error code like ngx.ERROR.
This Lua code execution context does support yielding, so Lua APIs that may yield (like cosockets, sleeping, and "light threads") are enabled in this context
Note, you need to configure the ssl_certificate and ssl_certificate_key to avoid the following error while starting NGINX:
nginx: [emerg] no ssl configured for the server
This directive requires OpenSSL 1.1.1 or greater.
If you are using the official pre-built packages for OpenResty 1.21.4.1 or later, then everything should work out of the box.
If you are not using the Nginx core shipped with OpenResty 1.21.4.1 or later, you will need to apply patches to the standard Nginx core:
https://openresty.org/en/nginx-ssl-patches.html
Note for HTTP/3 (QUIC) users: When using this directive with HTTP/3 connections, certain yield operations may fail if the QUIC SSL Lua yield patch is not applied to your OpenSSL installation. OpenResty packages include this patch by default, but if you are building lua-nginx-module separately, you may need to apply the patch manually to ensure proper yield/resume functionality for HTTP/3 connections in SSL Lua phases. The patch can be found at: nginx-1.27.1-quic_ssl_lua_yield.patch
This directive was first introduced in the v0.10.21 release.
syntax: ssl_client_hello_by_lua_file <path-to-lua-script-file>
context: http, server
phase: right-after-client-hello-message-was-processed
Equivalent to ssl_client_hello_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or, as from the v0.5.0rc32 release, the LuaJIT bytecode to be executed.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
Note for HTTP/3 (QUIC) users: When using this directive with HTTP/3 connections, certain yield operations may fail if the QUIC SSL Lua yield patch is not applied to your OpenSSL installation. OpenResty packages include this patch by default, but if you are building lua-nginx-module separately, you may need to apply the patch manually to ensure proper yield/resume functionality for HTTP/3 connections in SSL Lua phases. The patch can be found at: nginx-1.27.1-quic_ssl_lua_yield.patch
This directive was first introduced in the v0.10.21 release.
syntax: ssl_certificate_by_lua_block { lua-script }
context: server
phase: right-before-SSL-handshake
This directive runs user Lua code when Nginx is about to start the SSL handshake for the downstream SSL (https) connections.
It is particularly useful for setting the SSL certificate chain and the corresponding private key on a per-request basis. It is also useful to load such handshake configurations nonblockingly from the remote (for example, with the cosocket API). And one can also do per-request OCSP stapling handling in pure Lua here as well.
Another typical use case is to do SSL handshake traffic control nonblockingly in this context, with the help of the lua-resty-limit-traffic#readme library, for example.
One can also do interesting things with the SSL handshake requests from the client side, like rejecting old SSL clients using the SSLv3 protocol or even below selectively.
The ngx.ssl and ngx.ocsp Lua modules provided by the lua-resty-core library are particularly useful in this context. You can use the Lua API offered by these two Lua modules to manipulate the SSL certificate chain and private key for the current SSL connection being initiated.
This Lua handler does not run at all, however, when Nginx/OpenSSL successfully resumes the SSL session via SSL session IDs or TLS session tickets for the current SSL connection. In other words, this Lua handler only runs when Nginx has to initiate a full SSL handshake.
Below is a trivial example using the ngx.ssl module at the same time:
server {
listen 443 ssl;
server_name test.com;
ssl_certificate_by_lua_block {
print("About to initiate a new SSL handshake!")
}
location / {
root html;
}
}
See more complicated examples in the ngx.ssl and ngx.ocsp Lua modules' official documentation.
Uncaught Lua exceptions in the user Lua code immediately abort the current SSL session, so does the
ngx.exit call with an error code like ngx.ERROR.
This Lua code execution context does support yielding, so Lua APIs that may yield (like cosockets, sleeping, and "light threads") are enabled in this context.
Note, however, you still need to configure the ssl_certificate and ssl_certificate_key directives even though you will not use this static certificate and private key at all. This is because the NGINX core requires their appearance otherwise you are seeing the following error while starting NGINX:
nginx: [emerg] no ssl configured for the server
This directive requires OpenSSL 1.0.2e or greater.
If you are using the official pre-built packages for OpenResty 1.9.7.2 or later, then everything should work out of the box.
If you are not using the Nginx core shipped with OpenResty 1.9.7.2 or later, you will need to apply patches to the standard Nginx core:
https://openresty.org/en/nginx-ssl-patches.html
Note for HTTP/3 (QUIC) users: When using this directive with HTTP/3 connections, certain yield operations may fail if the QUIC SSL Lua yield patch is not applied to your OpenSSL installation. OpenResty packages include this patch by default, but if you are building lua-nginx-module separately, you may need to apply the patch manually to ensure proper yield/resume functionality for HTTP/3 connections in SSL Lua phases. The patch can be found at: nginx-1.27.1-quic_ssl_lua_yield.patch
This directive was first introduced in the v0.10.0 release.
syntax: ssl_certificate_by_lua_file <path-to-lua-script-file>
context: server
phase: right-before-SSL-handshake
Equivalent to ssl_certificate_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or, as from the v0.5.0rc32 release, the LuaJIT bytecode to be executed.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
Note for HTTP/3 (QUIC) users: When using this directive with HTTP/3 connections, certain yield operations may fail if the QUIC SSL Lua yield patch is not applied to your OpenSSL installation. OpenResty packages include this patch by default, but if you are building lua-nginx-module separately, you may need to apply the patch manually to ensure proper yield/resume functionality for HTTP/3 connections in SSL Lua phases. The patch can be found at: nginx-1.27.1-quic_ssl_lua_yield.patch
This directive was first introduced in the v0.10.0 release.
syntax: ssl_session_fetch_by_lua_block { lua-script }
context: http
phase: right-before-SSL-handshake
This directive runs Lua code to look up and load the SSL session (if any) according to the session ID provided by the current SSL handshake request for the downstream.
The Lua API for obtaining the current session ID and loading a cached SSL session data is provided in the ngx.ssl.session Lua module shipped with the lua-resty-core library.
Lua APIs that may yield, like ngx.sleep and cosockets, are enabled in this context.
This hook, together with the ssl_session_store_by_lua* hook, can be used to implement distributed caching mechanisms in pure Lua (based on the cosocket API, for example). If a cached SSL session is found and loaded into the current SSL connection context, SSL session resumption can then get immediately initiated and bypass the full SSL handshake process which is very expensive in terms of CPU time.
Please note that TLS session tickets are very different and it is the clients' responsibility to cache the SSL session state when session tickets are used. SSL session resumptions based on TLS session tickets would happen automatically without going through this hook (nor the ssl_session_store_by_lua* hook). This hook is mainly for older or less capable SSL clients that can only do SSL sessions by session IDs.
When ssl_certificate_by_lua* is specified at the same time, this hook usually runs before ssl_certificate_by_lua*. When the SSL session is found and successfully loaded for the current SSL connection, SSL session resumption will happen and thus bypass the ssl_certificate_by_lua* hook completely. In this case, Nginx also bypasses the ssl_session_store_by_lua* hook, for obvious reasons.
To easily test this hook locally with a modern web browser, you can temporarily put the following line in your https server block to disable the TLS session ticket support:
ssl_session_tickets off;
But do not forget to comment this line out before publishing your site to the world.
If you are using the official pre-built packages for OpenResty 1.11.2.1 or later, then everything should work out of the box.
If you are not using one of the OpenSSL packages provided by OpenResty, you will need to apply patches to OpenSSL in order to use this directive:
https://openresty.org/en/openssl-patches.html
Similarly, if you are not using the Nginx core shipped with OpenResty 1.11.2.1 or later, you will need to apply patches to the standard Nginx core:
https://openresty.org/en/nginx-ssl-patches.html
This directive was first introduced in the v0.10.6 release.
Note that this directive can only be used in the http context starting
with the v0.10.7 release since SSL session resumption happens
before server name dispatch.
syntax: ssl_session_fetch_by_lua_file <path-to-lua-script-file>
context: http
phase: right-before-SSL-handshake
Equivalent to ssl_session_fetch_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or rather, the LuaJIT bytecode to be executed.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
This directive was first introduced in the v0.10.6 release.
Note that: this directive is only allowed to used in http context from the v0.10.7 release
(because SSL session resumption happens before server name dispatch).
syntax: ssl_session_store_by_lua_block { lua-script }
context: http
phase: right-after-SSL-handshake
This directive runs Lua code to fetch and save the SSL session (if any) according to the session ID provided by the current SSL handshake request for the downstream. The saved or cached SSL session data can be used for future SSL connections to resume SSL sessions without going through the full SSL handshake process (which is very expensive in terms of CPU time).
Lua APIs that may yield, like ngx.sleep and cosockets,
are disabled in this context. You can still, however, use the ngx.timer.at API
to create 0-delay timers to save the SSL session data asynchronously to external services (like redis or memcached).
The Lua API for obtaining the current session ID and the associated session state data is provided in the ngx.ssl.session Lua module shipped with the lua-resty-core library.
To easily test this hook locally with a modern web browser, you can temporarily put the following line in your https server block to disable the TLS session ticket support:
ssl_session_tickets off;
But do not forget to comment this line out before publishing your site to the world.
This directive was first introduced in the v0.10.6 release.
Note that: this directive is only allowed to used in http context from the v0.10.7 release
(because SSL session resumption happens before server name dispatch).
syntax: ssl_session_store_by_lua_file <path-to-lua-script-file>
context: http
phase: right-after-SSL-handshake
Equivalent to ssl_session_store_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or rather, the LuaJIT bytecode to be executed.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
This directive was first introduced in the v0.10.6 release.
Note that: this directive is only allowed to used in http context from the v0.10.7 release
(because SSL session resumption happens before server name dispatch).
syntax: proxy_ssl_certificate_by_lua_block { lua-script }
context: location
phase: right-after-server-certificate-request-message-was-processed
This directive runs user Lua code when Nginx is about to post-process the SSL server certificate request message from upstream. It is particularly useful for setting the SSL certificate chain and the corresponding private key for the upstream SSL (https) connections. It is also useful to load such handshake configurations nonblockingly from the remote (for example, with the cosocket API).
The ngx.ssl.proxysslcert Lua module provided by the lua-resty-core library are particularly useful in this context.
Below is a trivial example using the ngx.ssl module and the ngx.ssl.proxysslcert module at the same time:
server {
listen 443 ssl;
server_name test.com;
ssl_certificate /path/to/cert.crt;
ssl_certificate_key /path/to/key.key;
location /t {
proxy_pass https://upstream;
proxy_ssl_certificate_by_lua_block {
local ssl = require "ngx.ssl"
local proxy_ssl_cert = require "ngx.ssl.proxysslcert"
-- NOTE: for illustration only, we don't handle error below
local f = assert(io.open("/path/to/cert.crt"))
local cert_data = f:read("*a")
f:close()
local cert, err = ssl.parse_pem_cert(cert_data)
local ok, err = proxy_ssl_cert.set_cert(cert)
local f = assert(io.open("/path/to/key.key"))
local pkey_data = f:read("*a")
f:close()
local pkey, err = ssl.parse_pem_priv_key(pkey_data)
local ok, err = proxy_ssl_cert.set_priv_key(pkey)
-- ...
}
}
...
}
See more information in the ngx.ssl.proxysslcert Lua module's official documentation.
Uncaught Lua exceptions in the user Lua code immediately abort the current SSL session, so does the
ngx.exit call with an error code like ngx.ERROR.
This Lua code execution context does support yielding, so Lua APIs that may yield (like cosockets, sleeping, and "light threads") are enabled in this context.
Note that, unlike the relations between the ssl_certificate and ssl_certificate_key directives and ssl_certificate_by_lua*, the proxy_ssl_certificate and proxy_ssl_certificate_key directives can be used together with proxy_ssl_certificate_by_lua*.
When there are only proxy_ssl_certificate and proxy_ssl_certificate_key directives, the original Nginx behavior will obviously remain the same.
When there is only proxy_ssl_certificate_by_lua*, Nginx will send the certificate and its related private key and chain set by Lua codes.
When the proxy_ssl_certificate and proxy_ssl_certificate_key directives and proxy_ssl_certificate_by_lua* are used at the same time, then proxy_ssl_certificate_by_lua* will take precedence over the proxy_ssl_certificate and proxy_ssl_certificate_key directives.
Please refer to corresponding test case file and ngx.ssl.proxysslcert for more details.
Note also that, it has the same condition as the proxy_ssl_certificate directive for proxy_ssl_certificate_by_lua* to work, that is the upstream server should enable verification of client certificates.
This directive requires OpenSSL 1.0.2e or greater.
syntax: proxy_ssl_certificate_by_lua_file <path-to-lua-script-file>
context: location
phase: right-after-server-certificate-request-message-was-processed
Equivalent to proxy_ssl_certificate_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or, as from the v0.5.0rc32 release, the LuaJIT bytecode to be executed.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
syntax: proxy_ssl_verify_by_lua_block { lua-script }
context: location
phase: right-after-server-certificate-message-was-processed
This directive runs user Lua code when Nginx is about to post-process the SSL server certificate message for the upstream SSL (https) connections.
It is particularly useful to parse upstream server certificate and do some custom operations in pure lua.
The ngx.ssl.proxysslverify Lua module provided by the lua-resty-core library are particularly useful in this context.
Below is a trivial example using the ngx.ssl.proxysslverify module at the same time:
server {
listen 443 ssl;
server_name test.com;
ssl_certificate /path/to/cert.crt;
ssl_certificate_key /path/to/key.key;
location /t {
proxy_ssl_certificate /path/to/cert.crt;
proxy_ssl_certificate_key /path/to/key.key;
proxy_pass https://upstream;
proxy_ssl_verify_by_lua_block {
local proxy_ssl_vfy = require "ngx.ssl.proxysslverify"
local cert = proxy_ssl_vfy.get_verify_cert()
-- ocsp to verify cert
-- check crl
proxy_ssl_vfy.set_verify_result()
...
}
}
...
}
See more information in the ngx.ssl.proxysslverify Lua module's official documentation.
Uncaught Lua exceptions in the user Lua code immediately abort the current SSL session, so does the
ngx.exit call with an error code like ngx.ERROR.
This Lua code execution context does support yielding, so Lua APIs that may yield (like cosockets, sleeping, and "light threads") are enabled in this context
This directive requires OpenSSL 3.0.2 or greater.
syntax: proxy_ssl_verify_by_lua_file <path-to-lua-script-file>
context: location
phase: right-after-server-certificate-message-was-processed
Equivalent to proxy_ssl_verify_by_lua_block, except that the file specified by <path-to-lua-script-file> contains the Lua code, or, as from the v0.5.0rc32 release, the LuaJIT bytecode to be executed.
When a relative path like foo/bar.lua is given, they will be turned into the absolute path relative to the server prefix path determined by the -p PATH command-line option while starting the Nginx server.
syntax: lua_shared_dict <name> <size>
default: no
context: http
phase: depends on usage
Declares a shared memory zone, <name>, to serve as storage for the shm based Lua dictionary ngx.shared.<name>.
Shared memory zones are always shared by all the Nginx worker processes in the current Nginx server instance.
The <size> argument accepts size units such as k and m:
http {
lua_shared_dict dogs 10m;
...
}
The hard-coded minimum size is 8KB while the practical minimum size depends on actual user data set (some people start with 12KB).
See ngx.shared.DICT for details.
This directive was first introduced in the v0.3.1rc22 release.
syntax: lua_socket_connect_timeout <time>
default: lua_socket_connect_timeout 60s
context: http, server, location
This directive controls the default timeout value used in TCP/unix-domain socket object's connect method and can be overridden by the settimeout or settimeouts methods.
The <time> argument can be an integer, with an optional time unit, like s (second), ms (millisecond), m (minute). The default time unit is s, i.e., "second". The default setting is 60s.
This directive was first introduced in the v0.5.0rc1 release.
syntax: lua_socket_send_timeout <time>
default: lua_socket_send_timeout 60s
context: http, server, location
Controls the default timeout value used in TCP/unix-domain socket object's send method and can be overridden by the settimeout or settimeouts methods.
The <time> argument can be an integer, with an optional time unit, like s (second), ms (millisecond), m (minute). The default time unit is s, i.e., "second". The default setting is 60s.
This directive was first introduced in the v0.5.0rc1 release.
syntax: lua_socket_send_lowat <size>
default: lua_socket_send_lowat 0
context: http, server, location
Controls the lowat (low water) value for the cosocket send buffer.
syntax: lua_socket_read_timeout <time>
default: lua_socket_read_timeout 60s
context: http, server, location
phase: depends on usage
This directive controls the default timeout value used in TCP/unix-domain socket object's receive method and iterator functions returned by the receiveuntil method. This setting can be overridden by the settimeout or settimeouts methods.
The <time> argument can be an integer, with an optional time unit, like s (second), ms (millisecond), m (minute). The default time unit is s, i.e., "second". The default setting is 60s.
This directive was first introduced in the v0.5.0rc1 release.
syntax: lua_socket_buffer_size <size>
default: lua_socket_buffer_size 4k/8k
context: http, server, location
Specifies the buffer size used by cosocket reading operations.
This buffer does not have to be that big to hold everything at the same time because cosocket supports 100% non-buffered reading and parsing. So even 1 byte buffer size should still work everywhere but the performance could be terrible.
This directive was first introduced in the v0.5.0rc1 release.
syntax: lua_socket_pool_size <size>
default: lua_socket_pool_size 30
context: http, server, location
Specifies the size limit (in terms of connection count) for every cosocket connection pool associated with every remote server (i.e., identified by either the host-port pair or the unix domain socket file path).
Default to 30 connections for every pool.
When the connection pool exceeds the available size limit, the least recently used (idle) connection already in the pool will be closed to make room for the current connection.
Note that the cosocket connection pool is per Nginx worker process rather than per Nginx server instance, so size limit specified here also applies to every single Nginx worker process.
This directive was first introduced in the v0.5.0rc1 release.
syntax: lua_socket_keepalive_timeout <time>
default: lua_socket_keepalive_timeout 60s
context: http, server, location
This directive controls the default maximal idle time of the connections in the cosocket built-in connection pool. When this timeout reaches, idle connections will be closed and removed from the pool. This setting can be overridden by cosocket objects' setkeepalive method.
The <time> argument can be an integer, with an optional time unit, like s (second), ms (millisecond), m (minute). The default time unit is s, i.e., "second". The default setting is 60s.
This directive was first introduced in the v0.5.0rc1 release.
syntax: lua_socket_log_errors on|off
default: lua_socket_log_errors on
context: http, server, location
This directive can be used to toggle error logging when a failure occurs for the TCP or UDP cosockets. If you are already doing proper error handling and logging in your Lua code, then it is recommended to turn this directive off to prevent data flushing in your Nginx error log files (which is usually rather expensive).
This directive was first introduced in the v0.5.13 release.
syntax: lua_ssl_ciphers <ciphers>
default: lua_ssl_ciphers DEFAULT
context: http, server, location
Specifies the enabled ciphers for requests to a SSL/TLS server in the tcpsock:sslhandshake method. The ciphers are specified in the format understood by the OpenSSL library.
The full list can be viewed using the “openssl ciphers” command.
This directive was first introduced in the v0.9.11 release.
syntax: lua_ssl_crl <file>
default: no
context: http, server, location
Specifies a file with revoked certificates (CRL) in the PEM format used to verify the certificate of the SSL/TLS server in the tcpsock:sslhandshake method.
This directive was first introduced in the v0.9.11 release.
syntax: lua_ssl_protocols [SSLv2] [SSLv3] [TLSv1] [TLSv1.1] [TLSv1.2] [TLSv1.3]
default: lua_ssl_protocols TLSv1 TLSv1.1 TLSv1.2 TLSv1.3
context: http, server, location
Enables the specified protocols for requests to a SSL/TLS server in the tcpsock:sslhandshake method.
The support for the TLSv1.3 parameter requires version v0.10.12 and OpenSSL 1.1.1.
From version v0.10.25, the default value change from SSLV3 TLSv1 TLSv1.1 TLSv1.2 to TLSv1 TLSv1.1 TLSv1.2 TLSv1.3.
This directive was first introduced in the v0.9.11 release.
syntax: lua_ssl_certificate <file>
default: none
context: http, server, location
Specifies the file path to the SSL/TLS certificate in PEM format used for the tcpsock:sslhandshake method.
This directive allows you to specify the SSL/TLS certificate that will be presented to server during the SSL/TLS handshake process.
This directive was first introduced in the v0.10.26 release.
See also lua_ssl_certificate_key and lua_ssl_verify_depth.
syntax: lua_ssl_certificate_key <file>
default: none
context: http, server, location
Specifies the file path to the private key associated with the SSL/TLS certificate used in the tcpsock:sslhandshake method.
This directive allows you to specify the private key file corresponding to the SSL/TLS certificate specified by lua_ssl_certificate. The private key should be in PEM format and must match the certificate.
This directive was first introduced in the v0.10.26 release.
See also lua_ssl_certificate and lua_ssl_verify_depth.
syntax: lua_ssl_trusted_certificate <file>
default: none
context: http, server, location
Specifies a file path with trusted CA certificates in the PEM format used to verify the certificate of the SSL/TLS server in the tcpsock:sslhandshake method.
This directive was first introduced in the v0.9.11 release.
See also lua_ssl_verify_depth.
syntax: lua_ssl_verify_depth <number>
default: lua_ssl_verify_depth 1
context: http, server, location
Sets the verification depth in the server certificates chain.
This directive was first introduced in the v0.9.11 release.
See also lua_ssl_certificate, lua_ssl_certificate_key and lua_ssl_trusted_certificate.
syntax: lua_ssl_key_log <file>
default: none
context: http, server, location
Enables logging of client connection SSL keys in the tcpsock:sslhandshake method and specifies the path to the key log file. Keys are logged in the SSLKEYLOGFILE format compatible with Wireshark.
syntax: lua_ssl_conf_command <command>
default: no
context: http, server, location
Sets arbitrary OpenSSL configuration commands.
The directive is supported when using OpenSSL 1.0.2 or higher and nginx 1.19.4 or higher. According to the specify command, higher OpenSSL version may be needed.
Several lua_ssl_conf_command directives can be specified on the same level:
lua_ssl_conf_command Options PrioritizeChaCha;
lua_ssl_conf_command Ciphersuites TLS_CHACHA20_POLY1305_SHA256;
Configuration commands are applied after OpenResty own configuration for SSL, so they can be used to override anything set by OpenResty.
Note though that configuring OpenSSL directly with lua_ssl_conf_command might result in a behaviour OpenResty does not expect, and should be done with care.
This directive was first introduced in the v0.10.21 release.
syntax: lua_upstream_skip_openssl_default_verify on|off
default: lua_upstream_skip_openssl_default_verify off
context: location, location-if
When using proxy_ssl_verify_by_lua directive, lua_upstream_skip_openssl_default_verify controls whether to skip default openssl's verify function, that means using pure Lua code to verify upstream server certificate.
This directive is turned off by default.
syntax: lua_http10_buffering on|off
default: lua_http10_buffering on
context: http, server, location, location-if
Enables or disables automatic response buffering for HTTP 1.0 (or older) requests. This buffering mechanism is mainly used for HTTP 1.0 keep-alive which relies on a proper Content-Length response header.
If the Lua code explicitly sets a Content-Length response header before sending the headers (either explicitly via ngx.send_headers or implicitly via the first ngx.say or ngx.print call), then the HTTP 1.0 response buffering will be disabled even when this directive is turned on.
To output very large response data in a streaming fashion (via the ngx.flush call, for example), this directive MUST be turned off to minimize memory usage.
This directive is turned on by default.
This directive was first introduced in the v0.5.0rc19 release.
syntax: rewrite_by_lua_no_postpone on|off
default: rewrite_by_lua_no_postpone off
context: http
Controls whether or not to disable postponing rewrite_by_lua* directives to run at the end of the rewrite request-processing phase. By default, this directive is turned off and the Lua code is postponed to run at the end of the rewrite phase.
This directive was first introduced in the v0.5.0rc29 release.
syntax: access_by_lua_no_postpone on|off
default: access_by_lua_no_postpone off
context: http
Controls whether or not to disable postponing access_by_lua* directives to run at the end of the access request-processing phase. By default, this directive is turned off and the Lua code is postponed to run at the end of the access phase.
This directive was first introduced in the v0.9.20 release.
syntax: precontent_by_lua_no_postpone on|off
default: precontent_by_lua_no_postpone off
context: http
Controls whether or not to disable postponing precontent_by_lua* directives to run at the end of the precontent request-processing phase. By default, this directive is turned off and the Lua code is postponed to run at the end of the precontent phase.
syntax: lua_transform_underscores_in_response_headers on|off
default: lua_transform_underscores_in_response_headers on
context: http, server, location, location-if
Controls whether to transform underscores (_) in the response header names specified in the ngx.header.HEADER API to hyphens (-).
This directive was first introduced in the v0.5.0rc32 release.
syntax: lua_check_client_abort on|off
default: lua_check_client_abort off
context: http, server, location, location-if
This directive controls whether to check for premature client connection abortion.
When this directive is on, the ngx_lua module will monitor the premature connection close event on the downstream connections and when there is such an event, it will call the user Lua function callback (registered by ngx.on_abort) or just stop and clean up all the Lua "light threads" running in the current request's request handler when there is no user callback function registered.
According to the current implementation, however, if the client closes the connection before the Lua code finishes reading the request body data via ngx.req.socket, then ngx_lua will neither stop all the running "light threads" nor call the user callback (if ngx.on_abort has been called). Instead, the reading operation on ngx.req.socket will just return the error message "client aborted" as the second return value (the first return value is surely nil).
When TCP keepalive is disabled, it is relying on the client side to close the socket gracefully (by sending a FIN packet or something like that). For (soft) real-time web applications, it is highly recommended to configure the TCP keepalive support in your system's TCP stack implementation in order to detect "half-open" TCP connections in time.
For example, on Linux, you can configure the standard listen directive in your nginx.conf file like this:
listen 80 so_keepalive=2s:2s:8;
On FreeBSD, you can only tune the system-wide configuration for TCP keepalive, for example:
# sysctl net.inet.tcp.keepintvl=2000
# sysctl net.inet.tcp.keepidle=2000
This directive was first introduced in the v0.7.4 release.
See also ngx.on_abort.
syntax: lua_max_pending_timers <count>
default: lua_max_pending_timers 1024
context: http
Controls the maximum number of pending timers allowed.
Pending timers are those timers that have not expired yet.
When exceeding this limit, the ngx.timer.at call will immediately return nil and the error string "too many pending timers".
This directive was first introduced in the v0.8.0 release.
syntax: lua_max_running_timers <count>
default: lua_max_running_timers 256
context: http
Controls the maximum number of "running timers" allowed.
Running timers are those timers whose user callback functions are still running or lightthreads spawned in callback functions are still running.
When exceeding this limit, Nginx will stop running the callbacks of newly expired timers and log an error message "N lua_max_running_timers are not enough" where "N" is the current value of this directive.
This directive was first introduced in the v0.8.0 release.
syntax: lua_sa_restart on|off
default: lua_sa_restart on
context: http
When enabled, this module will set the SA_RESTART flag on Nginx workers signal dispositions.
This allows Lua I/O primitives to not be interrupted by Nginx's handling of various signals.
This directive was first introduced in the v0.10.14 release.
syntax: lua_worker_thread_vm_pool_size <size>
default: lua_worker_thread_vm_pool_size 10
context: http
Specifies the size limit of the Lua VM pool (default 100) that will be used in the ngx.run_worker_thread API.
Also, it is not allowed to create Lua VMs that exceeds the pool size limit.
The Lua VM in the VM pool is used to execute Lua code in separate thread.
The pool is global at Nginx worker level. And it is used to reuse Lua VMs between requests.
Warning: Each worker thread uses a separate Lua VM and caches the Lua VM for reuse in subsequent operations. Configuring too many worker threads can result in consuming a lot of memory.
The various *_by_lua, *_by_lua_block and *_by_lua_file configuration directives serve as gateways to the Lua API within the nginx.conf file. The Nginx Lua API described below can only be called within the user Lua code run in the context of these configuration directives.
The API is exposed to Lua in the form of two standard packages ngx and ndk. These packages are in the default global scope within ngx_lua and are always available within ngx_lua directives.
The packages can be introduced into external Lua modules like this:
local say = ngx.say
local _M = {}
function _M.foo(a)
say(a)
end
return _M
Use of the package.seeall flag is strongly discouraged due to its various bad side-effects.
It is also possible to directly require the packages in external Lua modules:
local ngx = require "ngx"
local ndk = require "ndk"
The ability to require these packages was introduced in the v0.2.1rc19 release.
Network I/O operations in user code should only be done through the Nginx Lua API calls as the Nginx event loop may be blocked and performance drop off dramatically otherwise. Disk operations with relatively small amount of data can be done using the standard Lua io library but huge file reading and writing should be avoided wherever possible as they may block the Nginx process significantly. Delegating all network and disk I/O operations to Nginx's subrequests (via the ngx.location.capture method and similar) is strongly recommended for maximum performance.
syntax: val = ngx.arg[index]
context: set_by_lua*, body_filter_by_lua*
When this is used in the context of the set_by_lua* directives, this table is read-only and holds the input arguments to the config directives:
value = ngx.arg[n]
Here is an example
location /foo {
set $a 32;
set $b 56;
set_by_lua $sum
'return tonumber(ngx.arg[1]) + tonumber(ngx.arg[2])'
$a $b;
echo $sum;
}
that writes out 88, the sum of 32 and 56.
When this table is used in the context of body_filter_by_lua*, the first element holds the input data chunk to the output filter code and the second element holds the boolean flag for the "eof" flag indicating the end of the whole output data stream.
The data chunk and "eof" flag passed to the downstream Nginx output filters can also be overridden by assigning values directly to the corresponding table elements. When setting nil or an empty Lua string value to ngx.arg[1], no data chunk will be passed to the downstream Nginx output filters at all.
syntax: ngx.var.VAR_NAME
context: set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*, body_filter_by_lua*, log_by_lua*, balancer_by_lua*
Read and write Nginx variable values.
value = ngx.var.some_nginx_variable_name
ngx.var.some_nginx_variable_name = value
Note that only already defined Nginx variables can be written to. For example:
location /foo {
set $my_var ''; # this line is required to create $my_var at config time
content_by_lua_block {
ngx.var.my_var = 123
...
}
}
That is, Nginx variables cannot be created on-the-fly. Here is a list of pre-defined Nginx variables.
Some special Nginx variables like $args and $limit_rate can be assigned a value,
many others are not, like $query_string, $arg_PARAMETER, and $http_NAME.
Nginx regex group capturing variables $1, $2, $3, and etc, can be read by this
interface as well, by writing ngx.var[1], ngx.var[2], ngx.var[3], and etc.
Setting ngx.var.Foo to a nil value will unset the $Foo Nginx variable.
ngx.var.args = nil
CAUTION When reading from an Nginx variable, Nginx will allocate memory in the per-request memory pool which is freed only at request termination. So when you need to read from an Nginx variable repeatedly in your Lua code, cache the Nginx variable value to your own Lua variable, for example,
local val = ngx.var.some_var
--- use the val repeatedly later
to prevent (temporary) memory leaking within the current request's lifetime. Another way of caching the result is to use the ngx.ctx table.
Undefined Nginx variables are evaluated to nil while uninitialized (but defined) Nginx variables are evaluated to an empty Lua string.
This API requires a relatively expensive metamethod call and it is recommended to avoid using it on hot code paths.
context: init_by_lua*, set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*, body_filter_by_lua*, *log_by_lua*, ngx.timer.*, balancer_by_lua*, ssl_certificate_by_lua*, ssl_session_fetch_by_lua*, ssl_session_store_by_lua*, ssl_client_hello_by_lua*
ngx.OK (0)
ngx.ERROR (-1)
ngx.AGAIN (-2)
ngx.DONE (-4)
ngx.DECLINED (-5)
Note that only three of these constants are utilized by the Nginx API for Lua (i.e., ngx.exit accepts ngx.OK, ngx.ERROR, and ngx.DECLINED as input).
ngx.null
The ngx.null constant is a NULL light userdata usually used to represent nil values in Lua tables etc and is similar to the lua-cjson library's cjson.null constant. This constant was first introduced in the v0.5.0rc5 release.
The ngx.DECLINED constant was first introduced in the v0.5.0rc19 release.
context: init_by_lua*, set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*, body_filter_by_lua*, log_by_lua*, ngx.timer.*, balancer_by_lua*, ssl_certificate_by_lua*, ssl_session_fetch_by_lua*, ssl_session_store_by_lua*, ssl_client_hello_by_lua*
ngx.HTTP_GET
ngx.HTTP_HEAD
ngx.HTTP_PUT
ngx.HTTP_POST
ngx.HTTP_DELETE
ngx.HTTP_OPTIONS (added in the v0.5.0rc24 release)
ngx.HTTP_MKCOL (added in the v0.8.2 release)
ngx.HTTP_COPY (added in the v0.8.2 release)
ngx.HTTP_MOVE (added in the v0.8.2 release)
ngx.HTTP_PROPFIND (added in the v0.8.2 release)
ngx.HTTP_PROPPATCH (added in the v0.8.2 release)
ngx.HTTP_LOCK (added in the v0.8.2 release)
ngx.HTTP_UNLOCK (added in the v0.8.2 release)
ngx.HTTP_PATCH (added in the v0.8.2 release)
ngx.HTTP_TRACE (added in the v0.8.2 release)
These constants are usually used in ngx.location.capture and ngx.location.capture_multi method calls.
context: init_by_lua*, set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*, body_filter_by_lua*, log_by_lua*, ngx.timer.*, balancer_by_lua*, ssl_certificate_by_lua*, ssl_session_fetch_by_lua*, ssl_session_store_by_lua*, ssl_client_hello_by_lua*
value = ngx.HTTP_CONTINUE (100) (first added in the v0.9.20 release)
value = ngx.HTTP_SWITCHING_PROTOCOLS (101) (first added in the v0.9.20 release)
value = ngx.HTTP_OK (200)
value = ngx.HTTP_CREATED (201)
value = ngx.HTTP_ACCEPTED (202) (first added in the v0.9.20 release)
value = ngx.HTTP_NO_CONTENT (204) (first added in the v0.9.20 release)
value = ngx.HTTP_PARTIAL_CONTENT (206) (first added in the v0.9.20 release)
value = ngx.HTTP_SPECIAL_RESPONSE (300)
value = ngx.HTTP_MOVED_PERMANENTLY (301)
value = ngx.HTTP_MOVED_TEMPORARILY (302)
value = ngx.HTTP_SEE_OTHER (303)
value = ngx.HTTP_NOT_MODIFIED (304)
value = ngx.HTTP_TEMPORARY_REDIRECT (307) (first added in the v0.9.20 release)
value = ngx.HTTP_PERMANENT_REDIRECT (308)
value = ngx.HTTP_BAD_REQUEST (400)
value = ngx.HTTP_UNAUTHORIZED (401)
value = ngx.HTTP_PAYMENT_REQUIRED (402) (first added in the v0.9.20 release)
value = ngx.HTTP_FORBIDDEN (403)
value = ngx.HTTP_NOT_FOUND (404)
value = ngx.HTTP_NOT_ALLOWED (405)
value = ngx.HTTP_NOT_ACCEPTABLE (406) (first added in the v0.9.20 release)
value = ngx.HTTP_REQUEST_TIMEOUT (408) (first added in the v0.9.20 release)
value = ngx.HTTP_CONFLICT (409) (first added in the v0.9.20 release)
value = ngx.HTTP_GONE (410)
value = ngx.HTTP_UPGRADE_REQUIRED (426) (first added in the v0.9.20 release)
value = ngx.HTTP_TOO_MANY_REQUESTS (429) (first added in the v0.9.20 release)
value = ngx.HTTP_CLOSE (444) (first added in the v0.9.20 release)
value = ngx.HTTP_ILLEGAL (451) (first added in the v0.9.20 release)
value = ngx.HTTP_INTERNAL_SERVER_ERROR (500)
value = ngx.HTTP_NOT_IMPLEMENTED (501)
value = ngx.HTTP_METHOD_NOT_IMPLEMENTED (501) (kept for compatibility)
value = ngx.HTTP_BAD_GATEWAY (502) (first added in the v0.9.20 release)
value = ngx.HTTP_SERVICE_UNAVAILABLE (503)
value = ngx.HTTP_GATEWAY_TIMEOUT (504) (first added in the v0.3.1rc38 release)
value = ngx.HTTP_VERSION_NOT_SUPPORTED (505) (first added in the v0.9.20 release)
value = ngx.HTTP_INSUFFICIENT_STORAGE (507) (first added in the v0.9.20 release)
context: init_by_lua*, init_worker_by_lua*, set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*, body_filter_by_lua*, log_by_lua*, ngx.timer.*, balancer_by_lua*, ssl_certificate_by_lua*, ssl_session_fetch_by_lua*, ssl_session_store_by_lua*, exit_worker_by_lua*, ssl_client_hello_by_lua*
ngx.STDERR
ngx.EMERG
ngx.ALERT
ngx.CRIT
ngx.ERR
ngx.WARN
ngx.NOTICE
ngx.INFO
ngx.DEBUG
These constants are usually used by the ngx.log method.
syntax: print(...)
context: init_by_lua*, init_worker_by_lua*, set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*, body_filter_by_lua*, log_by_lua*, ngx.timer.*, balancer_by_lua*, ssl_certificate_by_lua*, ssl_session_fetch_by_lua*, ssl_session_store_by_lua*, exit_worker_by_lua*, ssl_client_hello_by_lua*
Writes argument values into the Nginx error.log file with the ngx.NOTICE log level.
It is equivalent to
ngx.log(ngx.NOTICE, ...)
Lua nil arguments are accepted and result in literal "nil" strings while Lua booleans result in literal "true" or "false" strings. And the ngx.null constant will yield the "null" string output.
There is a hard coded 2048 byte limitation on error message lengths in the Nginx core. This limit includes trailing newlines and leading time stamps. If the message size exceeds this limit, Nginx will truncate the message text accordingly. This limit can be manually modified by editing the NGX_MAX_ERROR_STR macro definition in the src/core/ngx_log.h file in the Nginx source tree.
context: init_worker_by_lua*, set_by_lua*, rewrite_by_lua*, access_by_lua*, content_by_lua*, header_filter_by_lua*, body_filter_by_lua*, log_by_lua*, ngx.timer.*, balancer_by_lua*, exit_worker_by_lua*
This table can be used to store per-request Lua context data and has a life time identical to the current request (as with the Nginx variables).
Consider the following example,
location /test {
rewrite_by_lua_block {
ngx.ctx.foo = 76
}
access_by_lua_block {
ngx.ctx.foo = ngx.ctx.foo + 3
}
content_by_lua_block {
ngx.say(ngx.ctx.foo)
}
}
Then GET /test will yield the output
79
That is, the ngx.ctx.foo entry persists across the rewrite, access, and content phases of a request.
Every request, including subrequests, has its own copy of the table. For example:
location /sub {
content_by_lua_block {
ngx.say("sub pre: ", ngx.ctx.blah)
ngx.ctx.blah = 32
ngx.say("sub post: ", ngx.ctx.blah)
}
}
location /main {
content_by_lua_block {
ngx.ctx.blah = 73
ngx.say("main pre: ", ngx.ctx.blah)
local res = ngx.location.capture("/sub")
ngx.print(res.body)
ngx.say("main post: ", ngx.ctx.blah)
}
}
Then GET /main will give the output
main pre: 73
sub pre: nil
sub post: 32
main post: 73
Here, modification of the ngx.ctx.blah entry in the subrequest does not affect the one in the parent request. This is because they have two separate versions of ngx.ctx.blah.
Internal redirects (triggered by nginx configuration directives like error_page, try_files, index, etc.) will destroy the original request ngx.ctx data (if any) and the new request will have an empty ngx.ctx table. For instance,
location /new {
content_by_lua_block {
ngx.say(ngx.ctx.foo)
}
}
location /orig {
content_by_lua_block {
ngx.ctx.foo = "hello"
ngx.exec("/new")
}
}
Then GET /orig will give
nil
rather than the original "hello" value.
Because HTTP request is created after SSL handshake, the ngx.ctx created
in ssl_certificate_by_lua*, ssl_session_store_by_lua*, ssl_session_fetch_by_lua* and ssl_client_hello_by_lua*
is not available in the following phases like rewrite_by_lua*.
Since v0.10.18, the ngx.ctx created during a SSL handshake
will be inherited by the requests which share the same TCP connection established by the handshake.
Note that overwrite values in ngx.ctx in the http request phases (like rewrite_by_lua*) will only take affect in the current http request.
Arbitrary data values, including Lua closures and nested tables, can be inserted into this "magic" table. It also allows the registration of custom meta methods.
Overriding ngx.ctx with a new Lua table is also supported, for example,
ngx.ctx = { foo = 32, bar = 54 }
When being used in the context of init_worker_by_lua*, this table just has the same lifetime of the current Lua handler.
The ngx.ctx lookup requires relatively expensive metamethod calls and it is much slower than explicitly passing per-request data along by your own function arguments. So do not abuse this API for saving your own function arguments because it usually has quite some performance impact.
Because of the metamethod magic, never "local" the ngx.ctx table outside your Lua function scope on the Lua module level due to worker-level data sharing. For example, the following is bad:
-- mymodule.lua
local _M = {}
-- the following line is bad since ngx.ctx is a per-request
-- data while this <code>ctx</code> variable is on the Lua module level
-- and thus is per-nginx-worker.
local ctx = ngx.ctx
function _M.main()
ctx.foo = "bar"
end
return _M
Use the following instead:
-- mymodule.lua
local _M = {}
function _M.main(ctx)
ctx.foo = "bar"
end
return _M
That is, let the caller pass the ctx table explicitly via a function argument.
syntax: res = ngx.location.capture(uri, options?)
context: rewrite_by_lua*, access_by_lua*, content_by_lua*
Issues a synchronous but still non-blocking Nginx Subrequest using uri.
Nginx's subrequests provide a powerful way to make non-blocking internal requests to other locations configured with disk file directory or any other Nginx C modules like ngx_proxy, ngx_fastcgi, ngx_memc,
ngx_postgres, ngx_drizzle, and even ngx_lua itself and etc etc etc.
Also note that subrequests just mimic the HTTP interface but there is no extra HTTP/TCP traffic nor IPC involved. Everything works internally, efficiently, on the C level.
Subrequests are completely different from HTTP 301/302 redirection (via ngx.redirect) and internal redirection (via ngx.exec).
You should always read the request body (by either calling ngx.req.read_body or configuring lua_need_request_body on) before initiating a subrequest.
This API function (as well as ngx.location.capture_multi) always buffers the whole response body of the subrequest in memory. Thus, you should use cosockets and streaming processing instead if you have to handle large subrequest responses.
Here is a basic example:
res = ngx.location.capture(uri)
Returns a Lua table with 4 slots: res.status, res.header, res.body, and res.truncated.
res.status holds the response status code for the subrequest response.
res.header holds all the response headers of the
subrequest and it is a normal Lua table. For multi-value response headers,
the value is a Lua (array) table that holds all the values in the order that
they appear. For instance, if the subrequest response headers contain the following
lines:
Set-Cookie: a=3
Set-Cookie: foo=bar
Set-Cookie: baz=blah
Then res.header["Set-Cookie"] will be evaluated to the table value
{"a=3", "foo=bar", "baz=blah"}.
res.body holds the subrequest's response body data, which might be truncated. You always need to check the res.truncated boolean flag to see if res.body contains truncated data. The data truncation here can only be caused by those unrecoverable errors in your subrequests like the cases that the remote end aborts the connection prematurely in the middle of the response body data stream or a read timeout happens when your subrequest is receiving the response body data from the remote.
URI query strings can be concatenated to URI itself, for instance,
res = ngx.location.capture('/foo/bar?a=3&b=4')
Named locations like @foo are not allowed due to a limitation in
the Nginx core. Use normal locations combined with the internal directive to
prepare internal-only locations.
An optional option table can be fed as the second argument, which supports the options:
method
specify the subrequest's request method, which only accepts constants like ngx.HTTP_POST.body
specify the subrequest's request body (string value only).args
specify the subrequest's URI query arguments (both string value and Lua tables are accepted)headers
specify the subrequest's request headers (Lua table only). this headers will override the original headers of the subrequest.ctx
specify a Lua table to be the ngx.ctx table for the subrequest. It can be the current request's ngx.ctx table, which effectively makes the parent and its subrequest to share exactly the same context table. This option was first introduced in the v0.3.1rc25 release.vars
take a Lua table which holds the values to set the specified Nginx variables in the subrequest as this option's value. This option was first introduced in the v0.3.1rc31 release.copy_all_vars
specify whether to copy over all the Nginx variable values of the current request to the subrequest in question. modifications of the Nginx variables in the subrequest will not affect the current (parent) request. This option was first introduced in the v0.3.1rc31 release.share_all_vars
specify whether to share all the Nginx variables of the subrequest with the current (parent) request. modifications of the Nginx variables in the subrequest will affect the current (parent) request. Enabling this option may lead to hard-to-debug issues due to bad side-effects and is considered bad and harmful. Only enable this option when you completely know what you are doing.always_forward_body
when set to true, the current (parent) request's request body will always be forwarded to the subrequest being created if the body option is not specified. The request body read by either ngx.req.read_body() or lua_need_request_body on will be directly forwarded to the subrequest without copying the whole request body data when creating the subrequest (no matter the request body data is buffered in memory buffers or temporary files). By default, this option is false and when the body option is not specified, the request body of the current (parent) request is only forwarded when the subrequest takes the PUT or POST request method.Issuing a POST subrequest, for example, can be done as follows
res = ngx.location.capture(
'/foo/bar',
{ method = ngx.HTTP_POST, body = 'hello, world' }
)
See HTTP method constants methods other than POST.
The method option is ngx.HTTP_GET by default.
The args option can specify extra URI arguments, for instance,
ngx.location.capture('/foo?a=1',
{ args = { b = 3, c = ':' } }
)
is equivalent to
ngx.location.capture('/foo?a=1&b=3&c=%3a')
that is, this method will escape argument keys and values according to URI rules and
concatenate them together into a complete query string. The format for the Lua table passed as the args argument is identical to the format used in the ngx.encode_args method.
The args option can also take plain query strings:
ngx.location.capture('/foo?a=1',
{ args = 'b=3&c=%3a' }
)
This is functionally identical to the previous examples.
The share_all_vars option controls whether to share Nginx variables among the current request and its subrequests.
If this option is set to true, then the current request and associated subrequests will share the same Nginx variable scope. Hence, changes to Nginx variables made by a subrequest will affect the current request.
Care should be taken in using this option as variable scope sharing can have unexpected side effects. The args, vars, or copy_all_vars options are generally preferable instead.
This option is set to false by default
location /other {
set $dog "$dog world";
echo "$uri dog: $dog";
}
location /lua {
set $dog 'hello';
content_by_lua_block {
res = ngx.location.capture("/other",
{ share_all_vars = true })
ngx.print(res.body)
ngx.say(ngx.var.uri, ": ", ngx.var.dog)
}
}
Accessing location /lua gives
/other dog: hello world
/lua: hello world
The copy_all_vars option provides a copy of the parent request's Nginx variables to subrequests when such subrequests are issued. Changes made to these variables by such subrequests will not affect the parent request or any other subrequests sharing the parent request's variables.
location /other {
set $dog "$dog world";
echo "$uri dog: $dog";
}
location /lua {
set $dog 'hello';
content_by_lua_block {
res = ngx.location.capture("/other",
{ copy_all_vars = true })
ngx.print(res.body)
ngx.say(ngx.var.uri, ": ", ngx.var.dog)
}
}
Request GET /lua will give the output
/other dog: hello world
/lua: hello
Note that if both share_all_vars and copy_all_vars are set to true, then share_all_vars takes precedence.
In addition to the two settings above, it is possible to specify
values for variables in the subrequest using the vars option. These
variables are set after the sharing or copying of variables has been
evaluated, and provides a more efficient method of passing specific
values to a subrequest over encoding them as URL arguments and
unescaping them in the Nginx config file.
location /other {
content_by_lua_block {
ngx.say("dog = ", ngx.var.dog)
ngx.say("cat = ", ngx.var.cat)
}
}
location /lua {
set $dog '';
set $cat '';
content_by_lua_block {
res = ngx.location.capture("/other",
{ vars = { dog = "hello", cat = 32 }})
ngx.print(res.body)
}
}
Accessing /lua will yield the output
dog = hello
cat = 32
The headers option can be used to specify the request headers for the subrequest. The value of this option should be a Lua table where the keys are the header names and the values are the header values. For example,
location /foo {
content_by_lua_block {
ngx.print(ngx.var.http_x_test)
}
}
location /lua {
content_by_lua_block {
local res = ngx.location.capture("/foo", {
headers = {
["X-Test"] = "aa",
}
})
ngx.print(res.body)
}
}
Accessing /lua will yield the output
aa
The ctx option can be used to specify a custom Lua table to serve as the ngx.ctx table for the subrequest.
location /sub {
content_by_lua_block {
ngx.ctx.foo = "bar";
}
}
location /lua {
content_by_lua_block {
local ctx = {}
res = ngx.location.capture("/sub", { ctx = ctx })
ngx.say(ctx.foo)
ngx.say(ngx.ctx.foo)
}
}
Then request GET /lua gives
bar
nil
It is also possible to use this ctx option to share the same ngx.ctx table between the current (parent) request and the subrequest:
location /sub {
content_by_lua_block {
ngx.ctx.foo = "bar"
}
}
location /lua {
content_by_lua_block {
res = ngx.location.capture("/sub", { ctx = ngx.ctx })
ngx.say(ngx.ctx.foo)
}
}
Request GET /lua yields the output
bar
Note that subrequests issued by ngx.location.capture inherit all the
request headers of the current request by default and that this may have unexpected side effects on the
subrequest responses. For example, when using the standard ngx_proxy module to serve
subrequests, an "Accept-Encoding: gzip" header in the main request may result
in gzipped responses that cannot be handled properly in Lua code. Original request headers should be ignored by setting
proxy_pass_request_headers to off in subrequest locations.
When the body option is not specified and the always_forward_body option is false (the default value), the POST and PUT subrequests will inherit the request bodies of the parent request (if any).
There is a hard-coded upper limit on the number of subrequests possible for every main request. In older versions of Nginx, the limit was 50 concurrent subrequests and in more recent versions, Nginx 1.9.5 onwards, the same limit is changed to limit the depth of recursive subrequests. When this limit is exceeded, the following error message is added to the error.log file:
[error] 13983#0: *1 subrequests cycle while processing "/uri"
The limit can be manually modified if required by editing the definition of the NGX_HTTP_MAX_SUBREQUESTS macro in the nginx/src/http/ngx_http_request.h file in the Nginx source tree.
Please also refer to restrictions on capturing locations configured by subrequest directives of other modules.
syntax: res1, res2, ... = ngx.location.capture_multi({ {uri, options?}, {uri, options?}, ... })
context: rewrite_by_lua*, access_by_lua*, content_by_lua*
Just like ngx.location.capture, but supports multiple subrequests running in parallel.
This function issues several parallel subrequests specified by the input table and returns their results in the same order. For example,
res1, res2, res3 = ngx.location.capture_multi{
{ "/foo", { args = "a=3&b=4" } },
{ "/bar" },
{ "/baz", { method = ngx.HTTP_POST, body = "hello" } },
}
if res1.status == ngx.HTTP_OK then
...
end
if res2.body == "BLAH" then
...
end
This function will not return until all the subrequests terminate. The total latency is the longest latency of the individual subrequests rather than the sum.
Lua tables can be used for both requests and responses when the number of subrequests to be issued is not known in advance:
-- construct the requests table
local reqs = {}
table.insert(reqs, { "/mysql" })
table.insert(reqs, { "/postgres" })
table.insert(reqs, { "/redis" })
table.insert(reqs, { "/memcached" })
-- issue all the requests at once and w
Truncated — view the full README on GitHub.
(top 30 of 150)
C
94.8%
Lua
3.2%
Shell
1.0%