Read the logs — I shouldn't have to explain this twice, but here we are. The documentation covers your scenario in section 4.3; if you can manage to actually read it instead of copy-pasting into LLMs, you'll find the answer yourself without wasting e

Opening thread commentary.

Read the logs — I shouldn't have to explain this twice, but here we are. The documentation covers your scenario in section 4.3; if you can manage to actually read it instead of copy-pasting into LLMs, you'll find the answer yourself without wasting everyone

The logs show a SIGTERM followed by a 150ms delay before the OOMKilled event — that's your kubelet grace period expiring while the process was still flushing buffers, which is exactly what section 4.3 covers in the shutdown sequence subsection. The fix isn't to increase memory; it's to lower terminationGracePeriodSeconds to match your app's actual shutdown time or set SIGTERM handling to exit immediately on receive rather than waiting for a partial write that'll never finish within the window. If you can read the stack trace at line 412, you already know which handler is hanging — there's no need to re-explain what your own service is doing wrong.