How we reduced 502 errors by caring about PID 1 in Kubernetes

The blog post discusses a technical issue encountered by GitLab's Site Reliability Engineering (SRE) team where GitLab Pages were consistently returning 502 errors, causing alert fatigue and impacting users. The problem was traced to GitLab Workhorse, a Go service that was not handling shutdown signals properly due to a misconfiguration in the Dockerfile CMD instruction, which used shell form instead of exec form. This oversight prevented GitLab Workhorse from terminating gracefully, leading Kubernetes to forcibly kill the process after 30 seconds, during which time 502 errors were generated. The team resolved the issue by updating the Dockerfile to ensure GitLab Workhorse received the termination signals correctly and implemented a progressive rollout with a feature flag to prevent disruption. The post highlights the importance of using the correct CMD syntax in Dockerfiles to manage processes effectively and prevent similar issues.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.