I am having a persistent redirection/configuration bug where I get the custom ruby-generated 404 page on my omnibus gitlab private instance instead of getting into the chat system, which has both the main gitlab and also a virtual host mapped for a chat (mattermost) server. My main site is gitlab.company.com and the chat site is gitlabchat.company.com. Something strange about nginx is that it seems that the virtual host “gitlabchat” which SHOULD work everywhere, only works reliably on Microsoft Edge and IE, and no longer works on Firefox 48.0.2 on Chrome 52.
Here’s a screenshot:
You can see that the site is “DOWN” in Chrome and “UP” and working in Edge. How can this even be? Is this an nginx bug? Is this some virtual hosting glitch? Obviously a single real IP is used for both my mattermost virtual host and the main gitlab virtual host, but the actual URL I am using I would have expected to work fine. I even wonder if my choice of “gitlabchat” (which I felt was more memorable to end users than mattermost, which is a horrifically stupid and non-memorable name) is the reason why I’m in 404 hell.
One thing I’m not sure of is whether I should be using an http or https URL in gitlab.rb when I configure the mattermost URL.
## Url on which GitLab will be reachable. ## For more details on configuring external_url see: ## https://gitlab.com/gitlab-org/omnibus-gitlab/blob/629def0a7a26e7c2326566f0758d4a27857b52a3/README.md#configuring-the-external-url-for-gitlab external_url 'http://gitlab.ramsoft.biz'
Changing the above to https brings the entire product screeching to a halt. I wonder why it’s so fragile. Nothing I can think of to fix this weird glitch or even debug it.
Update; I’m studying logs and my gitlab.rb and I think it’s a weird issue related to browser caching, browser security, and self-generated semi-bogus SSL certificates.