skydick/monitoring: drop InfluxDB zpool_io, add Prometheus pool-member map
The Grafana dashboards (skynet-web on door1) query Prometheus, not InfluxDB —
nothing queries skydick's InfluxDB — so the zpool_io telegraf collector fed
nothing. Replace it with a node-exporter textfile metric skyw_zpool_member{pool,
device}, refreshed from live `zpool status` each run. Grafana joins it onto
node_disk_*_completed_total to get honest *physical* per-pool IOPS, drift-proof
against kernel device renames (unlike a hardcoded device regex).

Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
1 parent 69a0199 commit b45bd98b29e8ff866766c7096c9349c6d669bd67
@Dixiao-L Dixiao-L authored 10 days ago
Showing 1 changed file
View
modules/monitoring.nix