A user, ToddZ, noticed a significant increase in Sidekiq job counts after updating to Discourse 3.5.0 on September 3, with the daily average rising from ~800 to ~1300 jobs. They also observed an unusual dip to 24 jobs on September 16, which they found puzzling. Another user, Jagster, suggested the trend might be normal when viewed over different timeframes. Falco explained that the increase could be due to more popular plugins being bundled with Discourse core, leading to more jobs without increased CPU usage. ToddZ accepted this explanation and decided not to worry about the change. ToddZ, Jagster, Falco
Self-hosting on test-passed branch here. Just noticed a couple of suprises in my Sidekiq graph this month:
Upon a Discourse update on Sep 3, the history graph jumped from a long-steady average of ~800/day to ~1300/day…
…with one very odd exception on Sep 16 when only 24 jobs were recorded. I wasn’t touching anything on the 16th, and I don’t see any unusual site activity.
Is there something new in the release I pulled on Sep 3 that causes many more jobs to run, or should I be concerned? Anything suspect about that Sep 16 dip?
The general trend continues, with a couple more oddities. Still wondering if anyone else observes a ~65% increase in Sidekiq jobs run since updating to 3.5.0. I’ve not noticed any performance changes, but I don’t know if those peaks and valleys are normal or concerning…
Depends on which was the previous version. It may be the Bundling more popular plugins with Discourse core and a lot of those are being early exited which would align with more jobs without more CPU usage.
Thanks, Falco. Sounds reasonable. I don’t have historical CPU data, but the site doesn’t feel any different. I’ll assume the bundling of plugins is behind the graph and not worry about it.
