we have a problem with Gitlab Pages.
When retrieving a page, we get the error message 502.
The log file (/var/log/gitlab/gitlab-pages/current)gives the following error message:
{"correlation_id":"01FFF3V9PQKTBHA50HF0XA3C6D","error":"Get \"OUR PAGES DOMAIN\": dial tcp XXX.XXX.XXX.XXX:443: connect: no route to host","host":"OUR DOMAIN","level":"error","msg":"could not fetch domain information from a source","path":"/favicon.ico","time":"2021-09-13T10:19:47+02:00"}
Here it is interesting that the IP XXX.XXX.XXX.XXX is not the IP of the server, it is an IP that is not assigned in out network.
We use version 14.2.3. The problems may have arisen when switching to version 14.
Update: We found a solution in our case. We use an older docker-compose.yml file, which contains a “hostname” attribute. We did set this do our fully qualified gitlab hostname, because we don’t know the impact and runs it for many years. We now changed this back to a neutral “gitlab”, which also works, because it only define linux hostname of docker container, but do not collide with our fqdn.
Old answer
I suppose you both are using SSL termination with an external nginx or traefik, correct?
In this case, docs propose to set nginx of docker image to only listen on http port.