Jobs seem often to stall at points unrelated to actual test execution

Here is an example, job #821403659 which stalled for 20 mins at the “fetch git repository” stage. I had to cancel and retry, it then ran fine. There was no indication there was any problem with the (private) runner machine, it seems to the the communication with the server that has stopped.

This happens at random, maybe one out of 30 jobs, and can also hit other stages like “Downloading artifacts”.

It’s been happening for a while (weeks) so is probably not related to the recent instabilities in the infrastructure.

If the above job# is not enough to investigate, I can find other failures to reference.

Just had it happen again for the artifacts below:

Downloading artifacts for behat_test (826881404)…