chore: use a smaller runner for e2e test #15034
Conversation
|
Ok. The e2e test is failing. Could it be because #14979 hasn't landed yet? |
There was a problem hiding this comment.
--max-workers=1
probably to reduce flakiness
Yes, at one point we had trouble running multiple tests in parallel because the machine was overloaded and started timing out. I think we could experiment to see if the situation has improved?
EDIT:
BTW if we want to increase parallelization level, we need to remember that this may affect API rate limit 👍
|
Thank you for the context @mtojek. We can investigate parallelism separately, my intention for this change now is to reduce CI compute cost. Also could you or someone from the team have a look why |
|
Using a 4 core 16GB runner solves the issue. This is interesting for me and I would like to kno why. |
We run our e2e-tests on a 16-core machine with
--max-workers=1Using a standard runner with 2 cores, the machine runs the tests in the same amount of time while reducing the cost 8 times.