We run Gitlab CE “on prem”, e.g. in our data centre, only accessible via our VPN.
For the most part it works well, although the UI is always slow.
I’m currently running a CI pipeline on one of our main codebase repositories, with some new Gitlab Runners. This pipeline has 1,000+ jobs, each of which take between 3 and 7 minutes (in parallel across many Runners).
The new Runners aren’t set up fully and are causing errors in some of the jobs.
The issue I’m having is… whilst this is happening, the UI is painfully slow. If I click on the pipeline in question, the UI hangs (page loads; AJAX content doesn’t) and sometimes seems to time out - e.g. it never loads. The same applies when I click into a job within the pipeline. Sometimes the page eventually loads, sometimes it doesn’t.
Two parts to this question really:
- Is this a bug? I can’t see any errors when I
gitlab-ctl tailthe logs
- Do I need to throw more resource (cores/RAM) at the Gitlab cluster to alleviate this? If so, what part of Gitlab (e.g. Sidekiq/Unicorn) is responsible for the UI and can I dedicate processes to it specifically?