raft: Add metrics monitoring performance of core Raft processing
The following metrics are designed to monitor the correctness, consistency, and performance of Gitaly. The context is that a storage can host hundreds of thousands of partitions. * gitaly_raft_proposal_duration_seconds Type: Histogram (exponential buckets) Labels: storage Description: Time to commit proposals (sampled) Purpose: Track write latency distribution * gitaly_raft_proposals_total Type: Counter Labels: storage, result (success/error) Description: Counter of all Raft proposals Purpose: Track overall throughput and error rates * gitaly_raft_log_entries_processed Type: Counter Labels: storage, operation (append/commit/apply), entry_type Description: Rate of log entries processed Purpose: Monitor overall system throughput
Showing
- internal/gitaly/storage/raftmgr/manager.go 43 additions, 2 deletionsinternal/gitaly/storage/raftmgr/manager.go
- internal/gitaly/storage/raftmgr/manager_test.go 239 additions, 21 deletionsinternal/gitaly/storage/raftmgr/manager_test.go
- internal/gitaly/storage/raftmgr/metrics.go 32 additions, 0 deletionsinternal/gitaly/storage/raftmgr/metrics.go
Please register or sign in to comment