Have you had a chance to take a look at 3.0.3?

This release includes a number of fixes and improvements in the 'Issue Navigator' area detailed in the following issues:

http://jira.atlassian.com/browse/JRA-5166
http://jira.atlassian.com/browse/JRA-4987

Let us know if the performance lag persists.

-Keith

Can you please turn on profiling (add '?jira_profile=on' to the URL, or else choose it from the admin section), and show us the profiling trace that comes to stardard out?

Apologies if you have sent this before - please point us to the appropriate email.

[Filter: profiling] Using parameter [jira_profile]
[Filter: profiling] defaulting to off [autostart=false]
[Filter: profiling] Turning filter off [jira_profile=off]

Did you change it through the administration section?

I will try turning on the system option now.

It runs to System out, as we didn't want to introduce any dependency on log4j into the profiling module. That way other people can use it (it is an open source project).

The profiling should go to the same place that the other output went:

[Filter: profiling] Using parameter [jira_profile]
[Filter: profiling] defaulting to off [autostart=false]
[Filter: profiling] Turning filter off [jira_profile=off]

The other option is to change the parameter in web.xml (it is an init param), but that would require a server restart.

Cheers,
Scott

I've just checked that all the std output indeed goes to catalina.out. I've also checked that Administration screen says profiling is turned on.

Still no profiling info is apparently printed either by default and after providing jira_profile URL parameter.

Your screenshot indicates that profiling is indeed turned on.

If you go to Issue Navigator while profiling is turned on and do a search you should see something like this appear in the logs:

PROFILED : 221 keys (4 unique) took 19ms/239ms : 7.949791% 0ms/query avg.
[1146ms] - /secure/IssueNavigator.jspa
[113ms] - IssueNavigator.execute()
[101ms] - IssueManager.execute()
[101ms] - Searching Issues
[19ms] - Lucene Query
[2ms] - Lucene Search
[5ms] - Retrieve From cache/db and filter
[1ms] - IssueManager.getIssue()
[1ms] - IssueManager.getIssue()
[1ms] - IssueManager.getIssue()
[1ms] - IssueManager.getIssue()
[77ms] - Sorting
[2ms] - Collection.iterator()
[1ms] - Collection.size()

Do these traces appear anywhere in the logs? Is JIRA updating the log, sometimes if the log is erroneously rotated by another active JIRA process the actual running version of JIRA does not have the reference to the log file anymore. In this case JIRA will need to be bounced, profiling turned on again and the search rerun.

Please let us knwo how you go.

Thanks,
Anton

The following JSP should help with debugging:

<%@ page import="com.atlassian.util.profiling.UtilTimerStack"%>
Writing to stdout.. <% System.out.println("To stdout"); %>
Writing to stderr.. <% System.err.println("To stderr"); %>
<% UtilTimerStack.push("Sleeping");
Thread.sleep(500);
UtilTimerStack.pop("Sleeping");
%>

I'm currently working on improving this. I'll get back to you with more information as I have it.

I'll ask you for your data sometime over the next few weeks to ensure that our performance improvements work for you.