Thread pool starvation triage checklist

Use this when CPU looks fine but the whole service starts timing out.

The download contains the real checklist and the logging schema. Start with thread-pool-starvation-triage-checklist.md.

What you get

• thread-pool-starvation-triage-checklist.md (the incident checklist)
• dependency-call-logging-schema.md (the log fields and examples that make queueing obvious)

When to use this

Use it when you see this cluster:

• latency rises across many endpoints at the same time
• throughput drops
• CPU is not pegged
• downstream timeouts appear (HTTP, SQL)
• recycle or restart appears to fix it temporarily

How to use it

• On call: run the checklist top to bottom, then capture one short artifact while it is slow.
• Tech lead: ship the smallest fixes that remove thread capture and add budgets (timeouts, concurrency caps, bounded retries).
• CTO: use the log schema to make risk measurable (timeouts by dependency, backlog shape, and repeat incidents).