Anton

Hope this helps,

David Fischer
Advestigo

Sorry about the poor English.
Thiago Rossato

1500 issues is quite a lot of new issues every day.
This error can also occur when you have some very large issues (1000's of comments). It can take quite a while to index these single issues and it can result in locks.
Make sure you are using the latest version of the Charting Plugin as this has fixed a few of the issues.

For 3.6 we are hopefully going to be introducing an indexing queue. This will stop the error in 99% of the cases.

Though can I ask how your issues are being created? Are they being imported from another system or do you have users actually creating 1500? We may be able to attach a patch that increases the time out for you. This may alleviate your problem.

Cheers,
Nick

These issues are created by about 800 users.
I really appreciate if you could send me a patch, because the reindexing process is tooking about 1 hour and 20 minutes.
And, to solve the problem above, just if we reindex, right?

By the way, right now, the reindex process didn´t work anymore.
I created an issue for this other problem (JSP-4409).

Please, I´m waiting for your answer very soon.
Imagine 800 users not working because of this error.

Thanks,
Thiago Rossato

Attached is an patch that will cause optimisations to happen half as often (every 300 issues), and uses a 30 second timeout.

Please let me know if this helps your problems.

Cheers,
Nick

2006-03-06 14:41:01,033 ERROR [jira.issue.index.DefaultIndexManager] Giving up reindex - waited 30 seconds
com.atlassian.jira.issue.index.IndexException
	at com.atlassian.jira.issue.index.DefaultIndexManager.getIndexLock(DefaultIndexManager.java:473)
	at com.atlassian.jira.issue.index.DefaultIndexManager.reIndexIssues(DefaultIndexManager.java:249)
	at com.atlassian.jira.issue.index.DefaultIndexManager.reIndexIssues(DefaultIndexManager.java:221)
	at com.atlassian.jira.issue.index.DefaultIndexManager.reIndex(DefaultIndexManager.java:401)
	at com.atlassian.jira.workflow.function.issue.IssueReindexFunction.execute(IssueReindexFunction.java:28)
	at com.opensymphony.workflow.AbstractWorkflow.executeFunction(AbstractWorkflow.java:1179)
	at com.opensymphony.workflow.AbstractWorkflow.transitionWorkflow(AbstractWorkflow.java:1434)
	at com.opensymphony.workflow.AbstractWorkflow.doAction(AbstractWorkflow.java:533)
	at com.atlassian.jira.workflow.SimpleWorkflowManager.doWorkflowAction(SimpleWorkflowManager.java:228)
	at com.atlassian.jira.workflow.WorkflowTransitionUtilImpl.progress(WorkflowTransitionUtilImpl.java:259)
	at com.atlassian.jira.web.action.issue.CommentAssignIssue.doExecute(CommentAssignIssue.java:188)
	at webwork.action.ActionSupport.execute(ActionSupport.java:153)

Is there an way to work without index optimizations until version 3.6 get ready?

The problem surfaces when there are 2 concurrent updates in the system, and at least one of them is in the middle of a transaction. If one thread obtains the index lock and then e.g. tries to retrieve comments for an issue, the db's query optimizer might decide that it is more efficient to do a table scan to retrieve the record. If another thread did an insert into the table, the select will block until the transaction that inserted a record is committed.

If the comment was added as part of the workflow transition, the transaction will not be committed until the issue is reindexed. However the reindex will not happen as the thread that it in the middle of the workflow operation (transaction) will try to obtain the indexing lock.

This causes a deadlock between the database and JIRA's indexing lock. It if often difficult to predict the behaviour of the query optimizer which causes the table scan and hence a deadlock. The solution is to perform indexing outside the database transaction.

This fix is really a good new for us!

Version 3.6.3 is under development. Is there a release date for this version?
Can you post a patch for this fix? You are experiencing many deadlock problems and indexes timeouts.

Thanks,
Thiago

We are hoping to release JIRA 3.6.3 some time next week.

I believe Nick has provided the patch for you in your open support case.

Thanks,
Anton

No patch was attached to my open support issue.
Can you ask someone to do this?

Yesterday our indexes got corrupted. I think that the problem reported by this issue is the reason.
Our company use JIRA 24x7. The reindex process took about 2 hours to be completed.

Cheers,
Thiago

Have you applied these?

cheers,
Nick

cheers,
parthi

The DefauiltIndexManager class being patched here is identical in 3.6.2 and 3.6.3, so the patch should work fine in either.

Thanks
Pierrick