until_and_while_executing is not running the job at all in sidekiq-unique-jobs 7.1.4 #626
Comments
|
It does say you have a duplicate, could it be that #624 caused this? It is the only thing I can think of. |
|
Exept removing all keys for 'uniquejobs:*' from Redis I also tried passing different arguments to worker |
|
Actually this scenario is misleading because in all honesty there is no lock so it shouldn't happen. I have seen someone mention this in the past but it should have been fixed and the only culprit it can be is #624. Unless it is one of those old unexplainable situations... Are you using redis namespace? |
|
I've tested sidekiq-unique-jobs 7.1.3 and it works fine. The bug was introduced in the 7.1.4 version. Before testing each version I cleared Redis keys with: r=REDIS_SIDEKIQ.keys('uniquejobs:*')
r.each { |k| REDIS_SIDEKIQ.del(k) } |
Yes. |
|
I can kind of simulate the problem. it "waits for brpoplpush when resolving the promise" do
allow(locksmith_one).to receive(:brpoplpush)
.with(anything, lock_timeout, &slow_brpoplpush)
.and_call_original
did_we_get_in = false
locksmith_one.execute do
did_we_get_in = true
end
expect(locksmith_one).to have_received(:brpoplpush)
expect(did_we_get_in).to be true
endThat works but now I have a warning saying:
There is no other stub of that method. No idea what is going on but I will restart my computer now :) |
|
I did some tests: sidekiq-unique-jobs 7.1.4
Output from Sidekiq: I downgraded sidekiq-unique-jobs 7.1.3
|
|
apologies for the regression @mhenrixon! What was the issue with passing in |
|
@millerjs not your fault, the issue was older than that. I (am the most likely scoundrel), had That was most likely the main issue, the code should have worked as it was and now the test you wrote is passing with a minimal change. Someone is still having a similar issue in the latest issue opened that jobs are just skipped so it seems there is still some minor adjustment to be done. |
Describe the bug
I've upgraded sidekiq-unique-jobs from 7.1.2 to 7.1.4 and I noticed that the worker with a strategy
until_and_while_executingdoes not process the job at all.Expected behavior
I expect the job to be processed.
Current behavior
When I run in rails console in the development environment:
Then I see in sidekiq logs:
I expect it should print
It works.Worker class
When I added
on_conflictthen I see raised error.Error raised for
on_conflict:Additional context
I was wondering that maybe there are some data left in Redis between sidekiq-unique-jobs 7.1.2 and 7.1.4 upgrade so I removed all keys:
It did not help. I still see the problem. The job content
puts It worksis not executed at all.The text was updated successfully, but these errors were encountered: