Performance degradation since upgrade to JIRA 8.2

[Paul Richard]

After upgrading to JIRA 8.2, it became next to unusable due to long waits in processing queries which returned multiple issues (e.g. filter, project queue, etc.). We tracked it down to the SQL + JQL Driver as JIRA is responsive again once disabled. I am now trying to nail down the cause.

I reviewed the prerequisites for the addon and have disabled gzip compression, but did not see any change. Regarding the memory requirement, does this mean I should be increasing the JVM's heap by an extra Gig? The VM on which JIRA is installed has a good 4GB free.

Any other suggestions? At first I thought it might be a badly written SQL statement that we'd used in one of our filters or elsewhere, but I'm not so sure about this now. Some of the slow queries I mentionned earlier aren't using custom SQL, just straight JQL. Nevertheless, is there a simple way to list out our custom uses of SQL (with this component) within JIRA?

Many thank!

[Pablo Beltran]

Hi Paul,

Might you share the SQL+JQL query slowing down your Jira and its execution plan?

(Type EXPLAIN before the query to see the execution plan)

Thanks,

Pablo

[Vimalraj]

Hi Paul,

You may check heap size and set it based on memory allocation. Another to check is garbage collection settings. I have tried both parallel and G1GC and prefer G1GC. 

Thanks,

Vimalraj

[Paul Richard]

Thanks Vimal, but it turns out that the problem was that we were using a SQL query within an SLA variable. I guess when running a filter or anything returning multiple issues, the SQL was called multiple times and slowed everything down. Since converting to using straight JQL (for these SLA variables), everything is snappier than ever. I can only assume that the plugin is now less efficient than it used to be perhaps though using SQL in an SLA variable appears to have been bad regardless as we now find JIRA generally faster than before the upgrade.