JIRA Behave Sluggish. Culprit?

[Tanner Wortham]

I'm looking through several links that help me diagnose performance issues, but while I do so, I wanted to ask the pros if what I describe below sounds familiar. Here's the details:

• JIRA behaves sluggish when updating issues, clicking on an issue to view it, refreshing an issue (hitting F5), advancing its workflow, commenting to an issue. Probably a few other update-related activities as well.
• By sluggish, I mean when you do any of the above, the browser just "spins" like it's waiting for the network to respond.
  • What's weird is it consistenty does so for exactly 11 seconds each and every time.
• JIRA runs super fast though when viewing a dashboard, creating an issue, deleting an issue.
• We just upgraded from 5.0.6 to 5.2.5. This is when I noticed the sluggishness.
• JIRA was just indexed a moment ago. Doesn't fix it.
• Below is a screenshot of a few things:
  • The "system info" screen of DEV.
  • A profiled log of an issue update and issue created found in the .txt file
• The profile on issue update turns up something suspect, but i'm not sure what it means:

2013-01-31 12:25:04,558 http-bio-8080-exec-3 DEBUG worthar 744x1223x1 lowvy6 10.210.75.249 /browse/MISSWHITE-224 [jira.web.filters.ThreadLocalQueryProfiler] com.atlassian.jira.config.properties.ApplicationProperties: 677 keys (6 unique) took 0ms/8518ms : 0.0% 0ms/query avg.

2013-01-31 12:25:04,558 http-bio-8080-exec-3 DEBUG worthar 744x1223x1 lowvy6 10.210.75.249 /browse/MISSWHITE-224 [jira.web.filters.ThreadLocalQueryProfiler]

2013-01-31 12:25:04,558 http-bio-8080-exec-3 DEBUG worthar 744x1223x1 lowvy6 10.210.75.249 /browse/MISSWHITE-224 [jira.web.filters.ThreadLocalQueryProfiler] RESULT GROUP: com.atlassian.jira.web.component.ModuleWebComponent

2013-01-31 12:25:04,558 http-bio-8080-exec-3 DEBUG worthar 744x1223x1 lowvy6 10.210.75.249 /browse/MISSWHITE-224 [jira.web.filters.ThreadLocalQueryProfiler] 2:8222ms renderModules [8207,15]

2013-01-31 12:25:04,558 http-bio-8080-exec-3 DEBUG worthar 744x1223x1 lowvy6 10.210.75.249 /browse/MISSWHITE-224 [jira.web.filters.ThreadLocalQueryProfiler] com.atlassian.jira.web.component.ModuleWebComponent: 2 keys (1 unique) took 8222ms/8518ms : 96.52501% 8222ms/query avg.

System Info:

Issue Create Profile:

(profile.txt)

[Tanner Wortham · Accepted solution]

i just happened to stumble across the root cause of our issue just a moment ago. at one point, we were evaluating crucible and fisheye. after evaluating, we decided to stick with what we had, we then took down the fisheye and crucible stuff, but forgot to remove its reference from the 'application links' section of JIRA. once i removed them, everything instantly sped up. i'm assuming that JIRA was reaching out to those modules, waited for a response, and ultimately timed out.

[Bjørn Olav Sagen]

We were experiencing the same problem as Tanner.

Removing the application links solved it.

[Harry Chan]

Hi, performance tuning is a very complicated exercise. If you want matters fixed, it is best to get an expert to go through every possibility and help tune it. You'd require testing to see if it'd break something (which can happen), if there are negative differences and reports produced on what's been changed for what reason.

This is vita in a production environment.

Having said that - to get you started, from your profile it looks like queries are slow. I'd suggest either your JIRA instance lacks memory (check via system information), there are performance issues with the database (JIRA has a tool to monitor this), there are not enough connections to the database (you can run thread dumps to verify) and if not many other possibilities which require further analysis.

[Adam Wride]

All of the sudden we're seeing a big drop off in performance. Not knowing where to start I would at least like to know what service is most constained (db, memory, etc). From this memory stats snapshot - would memory appear to be the culprit?http://d.pr/i/hQix