[SR-15353] tweak testSerialConcurrentPerform #13
Conversation
|
@swift-ci Please test Linux |
| }), | ||
| "Blocks didn't run serially" | ||
| ) | ||
| for (result, next) in zip(results, results.dropFirst()) { |
There was a problem hiding this comment.
I'd be really surprised if that changes anything. By the time this code runs all the results are already finalized - this code is just comparing them. I'd rather save the results into a log any time this assert fails on Linux and try to examine the logs to investigate why the test is failing. Just my 2￠
There was a problem hiding this comment.
I wasn't intending on fixing the test with this change, just have more visibility into what's failing and how. The actual fix will come once i can get CI to run it 😉
|
@swift-ci Please test Linux |
3 similar comments
|
@swift-ci Please test Linux |
|
@swift-ci Please test Linux |
|
@swift-ci Please test Linux |
|
@swift-ci Please test |
QuietMisdreavus
force-pushed
the
serial-concurrent
branch
from
98297ad
to
5090402
Compare
|
@swift-ci test |
|
It looks like the failure was because the loop was too fast, and caused some "start" timestamps to be identical to the previous "end" timestamp. I've updated the test to allow this; i'll run CI a few more times before marking the PR as ready. @swift-ci test |
|
@swift-ci test Linux |
|
@swift-ci test |
|
@swift-ci Please test |
|
@swift-ci test |
1 similar comment
|
@swift-ci test |
QuietMisdreavus
force-pushed
the
serial-concurrent
branch
from
44d0987
to
4788a67
Compare
|
@swift-ci test |
1 similar comment
|
@swift-ci test |
|
So we've now hit 10 test runs without an issue, so i think we're in the clear. I'll call this ready to go now. |
QuietMisdreavus
changed the title
Bug/issue #, if applicable: SR-15353
Summary
This PR is meant to help debug SR-15353 in CI, to figure out why the
testSerialConcurrentPerformtest is periodically failing on one of the Linux builders. Right now, it just tweaks the assertion logging, but i have a hunch as to what's going on (resolution/precision issues in thesystemUptimeclock) that i'll implement when i have a better picture of what's going wrong on the CI machine. (I couldn't reproduce it locally, whether in macOS or in a Linux Docker container.)Dependencies
N/A
Testing
This PR updates an existing test and doesn't touch any library code; behavior should not change.
Checklist
Make sure you check off the following items. If they cannot be completed, provide a reason.
./bin/testscript and it succeeded