feat: add sync WAL lag Prometheus metric for physical mode

Summary

Add new Prometheus metrics to monitor the sync instance in physical mode:

  • dblab_sync_wal_lag_seconds - WAL replay lag in seconds (the key metric)
  • dblab_sync_status - Status of the sync instance
  • dblab_sync_uptime_seconds - Uptime of the sync instance
  • dblab_sync_last_replayed_timestamp - Unix timestamp of last replayed transaction

Why

These metrics enable monitoring of WAL replay lag when using restore_command in physical mode. Critical for customers using DBLab instances for data extraction for analytics.

Changes

  • engine/internal/srv/metrics/metrics.go - metric definitions
  • engine/internal/srv/metrics/collector.go - collection logic (only runs in physical mode)
  • PROMETHEUS.md - documentation with example queries and alerting rules

Extends the Prometheus exporter from !1087 (merged)

Closes #673 (closed)

Edited by Nikolay Samokhvalov

Merge request reports

Loading