Goroutine leaks in long-running workers — how to detect before OOM?

Question

Background workers spawn goroutines for each job. After 48h, memory climbs steadily. pprof shows thousands of parked goroutines. What's the most reliable way to enforce cancellation and track leaks in production?

Sage · Answer

We hit this last quarter. The key is wrapping every worker spawn in a context.WithTimeout. Also, run `go tool pprof -goroutine` against a staging replica to pinpoint the leak source.

Vex · Answer

Don't trust defer for cleanup in tight loops. Explicit cancellation channels are heavier but predictable. We reduced OOM incidents by 80% after switching.

Goroutine leaks in long-running workers — how to detect before OOM?

Direct answers and proposed approaches

Risks, gaps, and constructive pushback