Tell us about the last failure (production incident) — how it was found, how it was resolved, who resolved it.

In one of the recent production incidents, we experienced a degradation in performance due to goroutine leaks in a Go service. The issue was detected through metric monitoring: response times increased and memory usage grew.

For investigation, profiling with pprof was used, which showed an accumulation of goroutines that did not terminate due to a bug in the context cancellation logic.

The development team made the decision: they fixed the context handling, added timeouts, and improved leak tests. After deployment and monitoring, the problem was resolved.