HyperIndex OpenInfra — Self-Hosted HyperIndex for Everyone

A free, self-hosted alternative to HyperIndex Cloud Offering that aims for 100% feature parity with @enviodev's dedicated plan.

You only need a HyperSync token, not a HyperIndex cloud plan.

Terraform brings up a Talos Linux Kubernetes cluster on EC2 — no EKS, no control-plane bill — and Flux converges everything else from this repo: Postgres, Hasura, the indexer, TLS, DNS, dashboards and alerts. Every tier feature below is a manifest you can read.

What you get

• Nothing is metered. Chains, contracts, storage, query rate, indexing hours. You pay EC2, not per hour indexed.
• Direct database access — -rw and -ro TLS endpoints straight at the CNPG pooler, each behind its own Envoy RBAC allowlist. Gated to the Dedicated plan upstream.
• Every alert the Production tier ships, routed to email, plus platform-level ones.
• A static GraphQL endpoint on your own DNS zone, Let's Encrypt certs renewed by cert-manager, records managed by external-dns.
• Backups you own — CNPG base backups and WAL archiving to your S3 bucket, on a schedule, with retention, replicated to a second region.
• Secrets from 1Password, pulled in by External Secrets and picked up on rotation by Reloader — no redeploy to change a variable.
• Grafana included, with dashboards for indexer health and the cluster underneath it.
• No lock-in. It is your AWS account, your cluster, your database.

Feature parity

✅ is in the tree, 🚧 is not yet, — is something that stops being a feature once the infrastructure is yours.

Alerts

Grafana alert rules, delivered by email. info is muted by default, so only warning and critical reach you. Every alert the Production tier offers:

The stack

flowchart LR
  HS["HyperSync"]
  CL["your clients"]

  subgraph gw["Envoy Gateway · deny by default"]
    R1["HTTPRoute<br/>hasura.your.zone"]
    R2["TLSRoute<br/>postgres-rw / -ro"]
  end

  subgraph k8s["Talos cluster on EC2"]
    IDX["indexer"]
    HAS["Hasura"]
    POOL["CNPG pooler"]
    PG[("Postgres<br/>3 instances")]
    OBS["Prometheus · Loki<br/>Grafana"]
  end

  S3[("S3")]
  S3R[("S3 replica<br/>second region")]

  HS -->|"events"| IDX
  CL -->|"GraphQL, IP allowlist"| R1
  CL -->|"psql, IP allowlist"| R2
  R1 --> HAS
  R2 --> POOL
  IDX -->|"writes rows"| PG
  HAS -->|"reads"| PG
  POOL --> PG
  PG -->|"base backups + WAL"| S3
  S3 -->|"cross-region replication"| S3R
  IDX -->|"metrics, logs"| OBS

Both ways in cross the same gateway, which denies anything not on the allowlist; the indexer and Hasura talk to Postgres directly, and only outside clients go through the pooler.

Layout

Getting started

Prerequisites, credentials and the four commands that stand a cluster up live in GUIDE.md.

Status

Built in the open, and running. Infrastructure, platform and the chain-indexer app are all in the tree; what is left is the 🚧 rows above — analytics over the indexed data, a ClickHouse deployment for the entities that want one, and a subgraph-compatible endpoint.