Our Self-hosted GitLab server is randomly getting broken and we couldn't figure out why. This random behavior affects our deployments, it gets too slow. After restarting it, stays up for a few hours and goes down, throws 500 or 502 error. After bringing it back up, I see either the sidekiq or gitaly metrics on omnibus grafana dashboard goes down compared to other services. What do these services do and how to debug this issue? Sidekiq metric image Gitaly metric image
System Info:
- OS - Centos 6.10 (final)
- CPU - 8 cores
- Memory usage - 8.82 GB / 15.6 GB
- Disk Usage -
- root : 111 GB / 389 GB
- boot : 169 MB / 476 MB
- /mnt/resource : 8.25 GB / 62.9 GB