Will decreasing the number of issues in a project improve JIRA performance?

[David Willox]

Currently in our PROD JIRA instance we have two projects with around 13,000 issues in one and another 30,000 in the other. Right now we see a search return of 2 seconds or under. We had recently experienced some indexing issues that caused "Lucene" errors in the application and dealt with those successfully by increasing the available resources to the application.

In the larger project we have recently re-categorized a number of the issue types and have discussed creating a new project to move those items to.
My question is, will there be any application benefits to moving these items? By widening the distribution of issues between 3 projects will there be any improvement in search speeds or background indexing?

Thanks,

David

[somethingblue]

Hi David,

The distribution of projects could help but the primary reason for performance would be the number of issues.  You don't have a lot of issues, howeer, if you never increased the heap memory that can have an affect since ~45,000 issues should have ~2GB of heap associated with it.

I would start by reviewing the JIRA Sizing guide to make sure you have the correct amount of heap and RAM allocated to the machine and the instance.  Once you confirm that is good, the next step would be to look in the JIRA_INSTALL/logs/catalina.out and the JIRA_HOME/log/atlassian-jira.log around the time you performed the search and see if there are any errors that stand out.

Finally, you can use the Troubleshooting Performance Issues with thread dumps knowledge base article to generate thread dumps and then use a site like fastthread.io to analyze the output.

Cheers,

Branden