Skip to content

Add more timeout happiness

Drew Blessing requested to merge dblessing/gitlab-ce:even_more_timeouts into master

@patricio and I spent more time troubleshooting LDAP with a customer today. Things go bad with LDAP, it's a reality. What makes it worse is when requests block indefinitely and legitimate requests pile up behind it. This adds even more timeout happiness to my previous merge request. Now all of our calls to LDAP, except for auth, will be subject to this timeout.

This also adds a retry mechanism. I stopped short of adding a configurable # of retries for now and just set it to retry once. I'm open to suggestions.

Things to check

I tried to reduce duplication here but one is a class method and one is instance. Can you see any way to reduce duplication?

Merge request reports