Skip to main content

Debug Production Issues

Scenario​

Users report slow response times. No alert has fired yet, but something is clearly wrong. You need to investigate quickly before it escalates.

How OpsWorker Helps​

Use AI Chat to interactively query your cluster:

You: Are there any pods with high restart counts in namespace production?

OpsWorker: Found 3 pods with elevated restart counts:
- api-gateway-7d8f9b (5 restarts in last hour)
- cache-worker-3k2m (3 restarts)
- payment-processor-9x1z (2 restarts)
You: Show me the logs for api-gateway. What errors are occurring?

OpsWorker: Recent error logs from api-gateway show:
- Connection timeout to redis-cache:6379 (repeated 47 times in last 30 min)
- "Failed to acquire connection from pool" errors starting at 14:23 UTC

If you have a GitHub or GitLab integration connected, the Source Code & Repository Agent can look at recent changes to the relevant manifests or app config:

You: What changed in the api-gateway deployment config recently?

OpsWorker (Source Code & Repository Agent): A change merged about 2 hours
ago reduced the Redis connection pool max from 50 to 10 in the
api-gateway configuration.

Root cause found: A recent config change reduced the Redis connection pool size, causing connection exhaustion under normal load.

Outcome​

  • Proactive debugging before an outage occurs
  • Identified root cause through interactive questioning, no need to chain kubectl commands
  • Connected the code change to infrastructure symptoms (requires a GitHub or GitLab integration)
  • Total time: about 3 minutes of conversation vs. 20+ minutes of manual investigation