-
Notifications
You must be signed in to change notification settings - Fork 141
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Improve flaky MG tests #807
Comments
this is a similar issue from a while ago: #377 we should check if it is still relevant, and if not, close it |
I'm starting some investigation on this. I notice MG is quite non-deterministic, so I think a good first step is to start logging here the different error messages we get with it. I'll do that in the comments below. |
from CI at ecb9c76 it went away after re-running the test
|
from my local setup at 96c593e I actually ran
|
from my local setup at 96c593e I also ran
|
from my local setup at 96c593e I also ran
|
@plebhash Good idea. Next thing should be add to the reports (maybe editing them so we don't lose context) the most likely cause that generate the error and a description of the error, and when possible a way to reproduce it. |
@pavlenex 2 things that will make us wait less are (1) split the MG tests in single units so that can be ran independently when they fail (2) use github cache for |
Another solution that I like less is to not require them in dev and fix the tests only before merging in main. So that we have to wait only one time instead of several times cause multiple PR in dev become only one in main |
concept NACK I think this one has potential to create chaos and confusion. If a PR to If we only see MG is broken when we are merging from Debugging MG is already quite complex, we should avoid introducing more complexity. |
reports is not possible, because if it breaks, no report is generated. |
I dont' get you point? Mayebe you misunderstood I mean editing your above messages |
yes I guess I misunderstood, I thought that by report you meant for the logs I pasted above, I already cleaned them up by following these criteria:
|
I believe this one should be fixed (hopefully by #905) If MG tests keep being unreliable we should log a new, more specified issue. |
Placeholder,until someone more knowledgable document sspecifics, but life shouldn't be waiting for MG tests to ✅ , so we should enure MG tests aren't flaky.
The text was updated successfully, but these errors were encountered: