Ever wanted to schedule jobs in Jira DC with more than 4 threads?

[Marty Henderson]

If you've ever fallen foul of the internal Jira scheduler limits like in https://jira.atlassian.com/browse/JRASERVER-65809 and https://jira.atlassian.com/browse/JRASERVER-70637 here's a way and a plugin that can help you increase the limits and take advantage of the hardware you're running on.

All the details as usual are here:  https://medium.com/@martyhenderson/configuring-jiras-scheduler-like-its-2024-ffad425cde53

The code for this is open source and highlights a simple way to actually fix this issue in Jira DC.  Let's hope this helps someone out!

Marty

[Ed Bukoski]

That is some seriously impressive work.  I almost wish I was still on DC so I could use this!

[Reece Lander _ScriptRunner - The Adaptavist Group_]

Sadly ScriptRunner still relies on Caesium / atlassian-scheduler.

This is a real pain when customers ask "why is my job not running?". The usual answer is mail handler jobs are blocking the thread pool. Other apps can also occupy all the scheduler threads. It's just a bad model to have N fixed threads in a shared system.

Virtual threads are probably a better solution, whenever Atlassian gets up to a modern JDK as a baseline.

Even if Atlassian just bumped the thread number up, that would be better than nothing...

If you have a long running task in an app, you should move it off onto another thread. This doesn't work if the scheduler never runs your job at all though.

I'd never ship a "hack" like this in ScriptRunner itself, having an app I can point customers to is useful.

One of these days Atlassian will either fix the issue, or I'll give up and bundle some alternative scheduler within ScriptRunner.

[Marty Henderson]

Hey Reece, totally agree with you on this. It’s a simple fix that could be provided, I’ve highlighted how it could be done and it sucks to have to resort to Java gymnastics to make it work. Let’s hope someone takes some notice of an easy fix.