monitoring: SAS SMART + ZFS pool textfile collectors for skydick
Closes the parity gap with door1 telegraf. node-exporter does not parse
SAS-specific smartctl output (predictive failure: grown defects, non-medium
errors, pending defects, ECC totals) — only SATA/NVMe attribute tables.
And the zfs collector exposes ARC + pool I/O but not pool health enum.

Adds skyw-textfile-collectors.service + .timer (5min cadence) that emits:
  smart_sas_power_on_hours{device}
  smart_sas_grown_defects{device}
  smart_sas_non_medium_errors{device}
  smart_sas_pending_defects{device}
  smart_sas_read_uncorrected{device}
  smart_sas_write_uncorrected{device}
  zpool_health{pool,state}     0=ONLINE 1=DEGRADED 2=FAULTED ...

Files chmod 0644 so node-exporter user can read them via the textfile
collector.

(Findings: sdd and sde on skydick already at 445 grown defects each.)
1 parent d52fca3 commit 110eaea633751a691979c43adb9f6f4f3f7a2df2
@ldx ldx authored on 9 May
Showing 1 changed file
View
modules/monitoring.nix