HASteward

High Availability, stewarded.

A Go CLI and Kubernetes operator that safely triages, repairs, backs up, and restores CNPG (PostgreSQL) and MariaDB/Galera clusters. Backups ride restic — block-level dedup, encryption, compression — and safety gates refuse to turn a bad day into a worse one.

Get Started CLI Reference View on GitHub
Latest version Docker pulls GitHub

  • Triage, read-only

    Diagnose a cluster's health — replicas, quorum, primary, lag — without touching a thing. Every repair starts here.

    Examples →

  • Repair, safely

    Heal unhealthy instances with a pre-repair escrow backup and refusal gates for ambiguous states. --dry-run shows the plan first.

    Safety Gates →

  • Backup & Restore

    restic-backed snapshots — block-level dedup, encryption, compression — with retention and point-in-time restore.

    Backups →

  • Operator Mode

    Run it in-cluster: CRDs, annotations, and a scheduler that triages and backs up on a cadence.

    Operator →

  • CNPG + Galera

    One steward for CloudNativePG (PostgreSQL) and MariaDB Operator (Galera), with engine-aware repair flows.

    Architecture →

  • Container-Native

    A single scratch image with the binary + restic. Drop it into a Job, a sidecar, or your shell.

    Container Usage →

Keep the cluster up. Let the steward handle the rest. → Browse the docs