viktor/infra
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

fix(monitoring): tune alert thresholds to reduce false positives

Browse source 
- HighPowerUsage: raise from 200W to 300W (R730 idles at ~230W)
- HighServiceLatency: exclude headscale (WebSocket) and authentik (SSO)
  from latency checks — both have inherently high avg response times
This commit is contained in:
Viktor Barzin 2026-04-06 15:39:23 +03:00
parent ca4acaecd0
commit b5689afe6d
1 changed files with 5 additions and 5 deletions
Download patch file Download diff file Expand all files Collapse all files

10
stacks/monitoring/modules/monitoring/prometheus_chart_values.tpl
View file

@ -756,12 +756,12 @@ serverFiles:
annotations:
summary: "Power outage - input voltage: {{ $value | printf \"%.0f\" }}V (threshold: <150V)"
- alert: HighPowerUsage
expr: r730_idrac_idrac_power_control_consumed_watts > 200
expr: r730_idrac_idrac_power_control_consumed_watts > 300
for: 60m
labels:
severity: info
annotations:
summary: "Server power: {{ $value | printf \"%.0f\" }}W (threshold: 200W)"
summary: "Server power: {{ $value | printf \"%.0f\" }}W (threshold: 300W)"
- alert: UsingInverterEnergyForTooLong
expr: automatic_transfer_switch_power_mode > 0 # 1 = Inverter; 0 = Grid
for: 24h
@ -1577,10 +1577,10 @@ serverFiles:
- alert: HighServiceLatency
expr: |
(
sum(rate(traefik_service_request_duration_seconds_sum{service!~".*idrac.*"}[5m])) by (service)
/ sum(rate(traefik_service_request_duration_seconds_count{service!~".*idrac.*"}[5m])) by (service)
sum(rate(traefik_service_request_duration_seconds_sum{service!~".*idrac.*|.*headscale.*|.*authentik.*"}[5m])) by (service)
/ sum(rate(traefik_service_request_duration_seconds_count{service!~".*idrac.*|.*headscale.*|.*authentik.*"}[5m])) by (service)
) > 10
and sum(rate(traefik_service_request_duration_seconds_count{service!~".*idrac.*"}[5m])) by (service) > 0.01
and sum(rate(traefik_service_request_duration_seconds_count{service!~".*idrac.*|.*headscale.*|.*authentik.*"}[5m])) by (service) > 0.01
and on() (time() - process_start_time_seconds{job="prometheus"}) > 900
for: 5m
labels: