// the find
signal18/replication-manager
Signal 18 repman - Replication Manager for MySQL / MariaDB / Percona Server
A full-featured HA orchestrator for MySQL/MariaDB/Percona replication topologies — does switchover, failover, proxy reconfiguration, backup scheduling, and cluster provisioning from a single binary. Aimed at DBAs and platform teams running serious MySQL infrastructure who don't want to glue together MHA, ProxySQL admin, and Restic themselves.
The proxy integration is genuinely useful — automated backend management for ProxySQL, MaxScale, and HAProxy means failover actually routes traffic, not just promotes a replica. The logplugin architecture is clever: pluggable Go binaries that analyze error logs, slow query logs, and binlogs for specific patterns (cleartext passwords in binlogs, InnoDB corruption, metadata lock contention) without baking everything into the core. Credential rotation with HashiCorp Vault integration is a real operational need that most similar tools ignore. The test suite has genuine integration tests covering backup concurrency, config diffs, and S3 provider behavior — not just unit tests with mocked everything.
730 stars for something this capable suggests it never broke out of the MariaDB-centric orbit — MySQL users will find rough edges and OpenSVC provisioning they've never heard of. The codebase has clear signs of accumulation: `cluster/` alone has 150+ files with no obvious domain boundaries, and the `doc/implementation/` tree is full of AI-generated session summaries masquerading as architecture docs, which tells you the contributors use Claude to write code and then dump the output into the repo. Configuration is a flat struct with hundreds of fields and multi-layer inheritance — functional but debugging 'why is this value set' will be painful. There's no clear upgrade path story; with this many moving parts (proxies, backup tools, provisioners), version compatibility between repman and the tools it orchestrates is underdocumented.