Symptom
The Cribl Stream Leader UI becomes slow or inaccessible while workers continue processing data. Jobs logs may show sustained event-loop utilization, heap failures, job timeouts, retained artifacts, or skipped collector runs.
Environment
- Cribl Stream distributed deployment
- Leader with scheduled collectors
- Jobs service responsible for collector scheduling and artifact management
Resolution
- Verify whether the data plane continues processing events.
- Verify Jobs-service event-loop utilization, heap, artifact counts, timeout counts, and skipped runs.
- Verify whether only the intended Leader is active.
- Identify collectors that run more frequently than their expected completion time.
- Reduce high-frequency schedules only after verifying state tracking and freshness requirements.
- Configure explicit timeouts for collectors that can hang on external dependencies.
- Review job-artifact retention and cleanup settings against current documentation.
- Review worker reconnect behavior after a Leader restart.
- Apply approved configuration changes during a controlled maintenance window.
- Verify sustained Jobs-service health and responsive UI pages after recovery.
Cause
This issue can occur when:
- Collector scheduling and artifact work exceeding Jobs-service capacity.
- Excessive completed-job or task-artifact retention.
- Collectors without effective timeouts occupying execution capacity.
- Simultaneous worker reconnects after a Leader restart.
- More than one active Leader scheduling the same collection workload.
- A product defect that causes retained state or RPC callbacks to grow.
Additional Information
A restart alone is not proof of recovery. If the same scheduling load and retained state return after restart, control-plane saturation can recur. Do not remove job-artifact files manually without an approved recovery procedure and backup plan.