Metrics — Time-Series Detail
Drill into any metric over any time range with full chart controls.
The Dashboard surfaces aggregate health at a glance. Metrics gives you the same time-series with full zoom, pan, multi-metric overlay, and percentile selection. This is where you go for incident retrospectives or capacity discussions, not first-pass triage.
What's available
Throughput: TPS commit, TPS rollback, queries per second by user. Latency: query mean, p95, p99 over time, broken out by query or aggregated. Connections: active, idle, idle-in-transaction; against the max_connections cap. I/O: buffer hit ratio, blocks read/written per second, WAL bytes per second. Cache: shared_buffers utilization, dirty page count. Replication: write/flush/replay lag in both bytes and time.
Controls
Time range picker top-right (15m / 1h / 6h / 24h / 7d / custom). Bucket size auto-adjusts based on the range; you can override from the gear menu if the auto-pick isn't what you want. Multi-metric overlay supports up to 4 series and gives each its own y-axis — combining commits and replication lag works, even though the units don't.
Export
The ⋯ menu offers PNG, SVG, or CSV. The CSV header has metric, server, bucket, and aggregation, so you can paste straight into a spreadsheet without renaming columns.