Files
git.stella-ops.org/docs/ops/scheduler-worker-prometheus-rules.yaml
Vladimir Moushkov 4d932cc1ba
Some checks failed
Docs CI / lint-and-preview (push) Has been cancelled
feat: Implement runner execution pipeline with planner dispatch and execution services
- Introduced RunnerBackgroundService to handle execution of runner segments.
- Added RunnerExecutionService for processing segments and aggregating results.
- Implemented PlannerQueueDispatchService to manage dispatching of planner messages.
- Created PlannerQueueDispatcherBackgroundService for leasing and processing planner queue messages.
- Developed ScannerReportClient for interacting with the scanner service.
- Enhanced observability with SchedulerWorkerMetrics for tracking planner and runner performance.
- Added comprehensive documentation for the new runner execution pipeline and observability metrics.
- Implemented event emission for rescan activity and scanner report readiness.
2025-10-27 18:57:35 +02:00

43 lines
1.9 KiB
YAML

groups:
- name: scheduler-worker
interval: 30s
rules:
- alert: SchedulerPlannerFailuresHigh
expr: sum(rate(scheduler_planner_runs_total{status="failed"}[5m]))
/
sum(rate(scheduler_planner_runs_total[5m])) > 0.05
for: 10m
labels:
severity: critical
service: scheduler-worker
annotations:
summary: "Planner failure ratio above 5%"
description: "More than 5% of planning runs are failing. Inspect scheduler logs and ImpactIndex connectivity before queues back up."
- alert: SchedulerPlannerLatencyHigh
expr: histogram_quantile(0.95, sum by (le) (rate(scheduler_planner_latency_seconds_bucket[5m]))) > 45
for: 10m
labels:
severity: warning
service: scheduler-worker
annotations:
summary: "Planner latency p95 above 45s"
description: "Planning latency p95 stayed above 45 seconds for 10 minutes. Check ImpactIndex, Mongo, or external selectors to prevent missed SLAs."
- alert: SchedulerRunnerBacklogGrowing
expr: max_over_time(scheduler_runner_backlog[15m]) > 500
for: 15m
labels:
severity: warning
service: scheduler-worker
annotations:
summary: "Runner backlog above 500 images"
description: "Runner backlog exceeded 500 images over the last 15 minutes. Verify runner workers, scanner availability, and rate limits."
- alert: SchedulerRunStuck
expr: sum(scheduler_runs_active) > 0 and max_over_time(scheduler_runs_active[30m]) == min_over_time(scheduler_runs_active[30m])
for: 30m
labels:
severity: warning
service: scheduler-worker
annotations:
summary: "Scheduler runs stuck without progress"
description: "Active runs count has remained flat for 30 minutes. Investigate stuck segments or scanner timeouts."