master is green, but deployment is not functional after sidekiq & unicorn merge
We noticed after merging !11 (merged), !16 (merged), and !19 (merged) that while the master pipeline was
Problems
- Unicorn had missing properties in
.gitlab-ci.yml
-
gitlab_url
which should be the FQDN of theexternal_url
-
- Sidekiq: the keys for
redis
were mis-named. - Changes need made to the destination for the
gitlab-*
url in NGINX configuration properties - The
unicorn
pod was not restarted when the Unicorn process had failed, due to the method used inprocess-wrapper
inside the contaner (#69 (closed)) - The
omnibus
chart probes were not functional, because affected services were not running, causing it to continually restart, resulting in Unicorn death as PSQL was not alive.
Steps forward
-
#69 (closed) @twk3-p4-open-source -
Remove omnibus
chart probes @WarheadsSE -
Add gitlab-healthcheck
probes fromkubernetes-gitlab-demo
chart(s) -
Add missing properties to helm --set
@WarheadsSE
Edited by Marin Jankovski