fix: extend wait timeout for test_delete_user() #1316
Conversation
Codecov Report
@@ Coverage Diff @@
## master #1316 +/- ##
=======================================
Coverage 80.76% 80.76%
=======================================
Files 69 69
Lines 3623 3623
=======================================
Hits 2926 2926
Misses 697 697
Flags with carried forward coverage won't be shown. Click here to find out more. Continue to review full report at Codecov.
|
|
Thanks for looking into this one. TBH I'm not sure anymore if the sidekiq metrics are a reliable source for the slow async operations - the point was to avoid hardcoded sleeps due to the underpowered CI VMs. I haven't taken the time to investigate it, so if you can solve it that's great, but don't lose too much time over this :) |
Thanks. Doing some debugging at the moment. I'll see what I figure out. |
|
Not sure if this fixes the issue. Especially if the sidekick operations can not be counted on. But it should not hurt. |
Have been seeing intermittent failures of the test_delete_user() functional test. Have made the following changes to hopefully resolve the issue and if it still fails to know better why the failure occurred. * Extend the wait timeout for test_delete_user() from 30 to 60 tries of 0.5 seconds each. * Modify wait_for_sidekiq() to return True if sidekiq process terminated. Return False if the timeout expired. * Modify wait_for_sidekiq() to loop through all processes instead of assuming there is only one process. If all processes are not busy then return. * Modify wait_for_sidekiq() to sleep at least once before checking for processes being busy. * Check for True being returned in test_delete_user() call to wait_for_sidekiq()
Cool, I guess the only thing that might happen is that some rogue sidekiq processes will drag this longer, but this was probably already the case and I forget what I was trying to achieve with relying only on the first process 😁 |
No description provided.