Scripted field causes Issue navigator slow loading - how to improve performance?

Jiri Vomacka

I use Jira 6.1 with latest Script Runner, and I use (a quite simple) scripted field computing total time of linked issues (in Epic):

import com.atlassian.jira.ComponentManager
double totalTime = 0
def scripttime=System.currentTimeMillis()

if (issue.getIssueTypeId()=="21") { // Epic
  def componentManager = ComponentManager.getInstance()
  def issueLinkManager = componentManager.getIssueLinkManager()
  def time=0
  issueLinkManager.getOutwardLinks(issue.id)?.each {issueLink ->
    if (issueLink.issueLinkType.name == "Epic-Story Link") {
        time=issueLink.destinationObject.getTimeSpent();
        if (time!=null) totalTime += time/3600;        
        }
    }
  }     
scripttime = System.currentTimeMillis()-scripttime;
log.warn ("ScriptField Epic time summary took [ms] "+scripttime)
return totalTime

It works well on issue detail, but when I use this field in Issue navigator, JIRA whows the results extremely slowly.
The script itself takes only several milliseconds (from 1 to 10), but the Jira profiler says that each item takes between 150 and 500 milliseconds. Result: issue navigator with this field displayed and JQL query with 100 items takes

approx. 25 seconds to load and display.

JIRA evaluates the custom field for each issue in the navigator - is there some overhead in Jira, while calling and executing groovy script?

I would expected that Jira takes the values from Lucene index...
Custom field has "Number searcher" assigned.

Please, could you help me with this problem?

JamieA

I can't reproduce any problem here at all. I tried your script, around 100 issues, displaying the script field value in the issue navigator, and it takes around 500ms total.

There must be some other problem. What is the template for this field?

> JIRA evaluates the custom field for each issue in the navigator - is there some overhead in Jira, while calling and executing groovy script?

Not in the sense that you mean... it's not treated as a script, it's just calling a method on a class that's in memory. The script is only compiled to a class once.

> I would expected that Jira takes the values from Lucene index...

It uses the index for searching, but not for displaying the values. I agree with you though, that would be better.

Rather than checking for issuetype == 21, why don't you associate this field only with Epics... at least that will mean it will only be evaluated for epics, and you can see if that makes a difference.

But like I say, for me, a hundred invocations of this script takes < 10ms.

Also try replacing the script with just "return 1", and see if that makes a difference. Try to narrow this down a bit...