I just stumbled upon this build, where CI failed to start an Elasticsearch containers. See below for the logs.

It’s apparently a timeout, and it doesn’t happen often. I think Elasticsearch is just slow to start on CI.

A quick google search didn’t turn anything useful, so we’ll need to investigate a bit more.

I can just say that the bit about Unable to load config file null doesn’t seem to be critical (even if it doesn’t seem normal), as the boot continues afterwards.

Also, if I try to start the container locally, there’s a lot of templates being loaded on startup over 1-2s, so I suppose it could be 5-10s on a slower machine. We might want to disable a few things to get back to a decently fast startup.