Pulse running Job Cleanup scan, but option disabled

Deadline
1

We are (still) having memory issues with Pulse that lead to periodic crashing. I am planning to upgrade to 7.1 as soon as I test it and have the opportunity, but that’s a non-trivial process to execute across two studios in the middle of production. But I digress…

I’ve noticed from looking at the Pulse log that it is still performing a Job Cleanup Scan, in spite of the fact that the option to automatically clean up jobs is disabled in the repository settings. I don’t think the memory situation is being helped by the fact that the cleanup scan is currently loading over 1,600 jobs (we have around 2,100 in the queue in question right now), which takes over 10 seconds (so there’s some heavyweight data being read). I tried bumping the job cleanup threshold from 0 to 128 hours (leaving the cleanup option itself disabled), on the off-chance that Pulse was using this value to query Mongo more intelligently, but that didn’t change the number of jobs being loaded.

So, my question is, shouldn’t Pulse just skip the Job Cleanup scan altogether if the option is disabled in the repo?

2

Hello Nathan,

We are going to need to test this internally. Can you verify your version is 7.0.3.0? Thanks.

3

We’re actually on 7.0.2.3.

4

Hello Nathan,

I have been able to verify what you are seeing, and when I talked to the devs, they advised that the job cleanup scan needs to load all the completed jobs to check if any of them need to be archived or deleted based on their On Job Complete setting. This is a per-job setting which needs to be checked for each job, regardless of whether or not automatic job cleanup is enabled in the repository options. Therefore it looks like this is by design, and not a bug.