-
Notifications
You must be signed in to change notification settings - Fork 29.7k
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
test: fix tick-processor tests and run them in CI #23100
Conversation
Note that the existing tests do not catch #22825 (comment). |
0233517
to
f9fdd79
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
Needs a rebase. |
ping @BridgeAR 🔉 |
These tests were removed from the main suite because:
If we could make sure these are resolved before moving to the CI suite, that would be 👍. |
That is pretty much exactly what I tried to achieve here. It is still difficult to guarantee that the tests are not flaky though. So far they should not leave any extra processes running. The flakiness is significantly reduced but not yet completely fixed. I am working on that. |
This reduces the runtime significantly and fixes the tests to run on the CI. It is significantly more robust by accepting some fallbacks. Therefore they are moved out of pummel to sequential.
f9fdd79
to
80dc417
Compare
Those tests were doing significantly to much work and almost all failed for me before with timeouts after 2 minutes. The test was super heavy and there was a lot of IO involved. I reduced the runtime of each of these tests to about 1-3 seconds (depending on finding a match right away or if they have to try again) and moved them to sequential, so we would actually know when they fail.
I removed one of the tests as it seems that the V8 internals do not produce "UNKNOWN" frequently anymore and it was already very rare when the test was written. In the end these tests are mainly redundant to each other anyway, as it should be enough to have a single test to check for the failure and another one to check for a specific output. Checking for specific output with specific code should not be required but I thought I keep the two redundant ones anyway.
This is still WIP as I want to see if they actually still fail on a lot of OS or if we are able to run them on most systems by now.
Checklist
make -j4 test
(UNIX), orvcbuild test
(Windows) passes