This story was originally published on HackerNoon at:
https://hackernoon.com/debugging-intermittent-kong-503s-when-the-logs-showed-nothing-but-the-status-code.
Investigating intermittent Kong 503 errors with quiet logs, tracing the failure to an NGINX worker restart caused by Kubernetes memory pressure.
Check more stories related to programming at:
https://hackernoon.com/c/programming.
You can also check exclusive content about
#kong-gateway,
#kong,
#nginx,
#nginx-worker-restart,
#kubernetes-oom,
#lua-vm-metrics,
#api-gateway-debugging,
#sre-incident-analysis, and more.
This story was written by:
@gudevamsikrishna. Learn more about this writer by checking
@gudevamsikrishna's about page,
and for more stories, please visit
hackernoon.com.
We investigated intermittent Kong 503 responses where the logs showed only the status code and an empty upstream address. Worker-level metrics and dmesg eventually showed that a cgroup OOM event had killed one of the two NGINX workers in the Kong pod.