High CPU, lot of long running threads on lucene indexes in JIRA 6.0.8

[Sridhar Nimmagadda]

Hi,

I have a situation where my server CPU utilization starts increasing and stays on for a very long time (haven't measured exactly how long, i have see it high for about 1-2 hrs)

JIRA becomes unresponsive and everything seems to fall back to normal upon restart

I took thread dumps(20 every 3 seconds ) during the hign CPU time and i get the below stack for all the long running threads

I am not able to figureout, why high CPU utilization stays on for a very long time and comes back to normal after restart.

Most of these long running threads originate from , JIRA Ajile, gadgets, Script Runner other JQLs

at sun.misc.Unsafe.getByte(Native Method) at java.nio.DirectByteBuffer.get(Unknown Source) at java.nio.ByteBuffer.get(Unknown Source) at java.nio.DirectByteBuffer.get(Unknown Source) at org.apache.lucene.store.MMapDirectory$MMapIndexInput.readBytes(MMapDirectory.java:289) at org.apache.lucene.index.TermBuffer.read(TermBuffer.java:82) at org.apache.lucene.index.SegmentTermEnum.next(SegmentTermEnum.java:131) at com.atlassian.jira.issue.search.parameters.lucene.sort.JiraLuceneFieldFinder.getMatches(JiraLuceneFieldFinder.java:232) at com.atlassian.jira.issue.search.parameters.lucene.sort.JiraLuceneFieldFinder.getUniqueMatches(JiraLuceneFieldFinder.java:184) at com.atlassian.jira.issue.statistics.util.OneDimensionalObjectHitCollector.setNextReader(OneDimensionalObjectHitCollector.java:92) at org.apache.lucene.search.IndexSearcher.search(IndexSearcher.java:531) at org.apache.lucene.search.IndexSearcher.search(IndexSearcher.java:320) at com.atlassian.jira.index.DelegateSearcher.search(DelegateSearcher.java:153)

thanks in advanace.

[NarendranB]

We are facing same issue. Any fixes available to try?

[Sridhar Nimmagadda]

Can you please elobarate the issue you are facing? do you have the same thread pile up of lucene indexes ?

[NarendranB]

We have a powerful server and in spite of that almost every day, the CPU load increases and JIRA becomes unresponsive. Therad dumps reveal that threads piling up on Lucene search. We are on JIRA 6.0.6. I see that you faced this problem back in May. Are you still facing this? If not, did you try any fixes?

[Sridhar Nimmagadda]

Can you compare the thread dump when there is high CPU and when the CPU is OK and post the threads that are piling up

[codelab expert]

Hi Sridhar,

you can try to rebuild your indexes from ground. You'll find the indexes in this folder \DATA\caches\indexes\.

1. Shutdown JIRA

2. Delete indexes

3. Start JIRA

4. Go to administration panel and recreate indexes

I hope this will help you!

[Sridhar Nimmagadda]

That didn't work..

I have almost 90 threads with the below stack(cut off the remaining stack), which is using up most (> 80%)of the CPU, all in runnable state

at org.apache.lucene.store.DataInput.readVInt(DataInput.java:105)
at org.apache.lucene.index.SegmentTermDocs.read(SegmentTermDocs.java:144)
at org.apache.lucene.search.TermScorer.nextDoc(TermScorer.java:112)
at org.apache.lucene.search.ConjunctionScorer.nextDoc(ConjunctionScorer.java:127)
at org.apache.lucene.search.BooleanScorer2.nextDoc(BooleanScorer2.java:303)
at org.apache.lucene.util.OpenBitSetDISI.inPlaceOr(OpenBitSetDISI.java:52)
at org.apache.lucene.util.OpenBitSetDISI.(OpenBitSetDISI.java:32)
at org.apache.lucene.search.CachingWrapperFilter.docIdSetToCache(CachingWrapperFilter.java:181)
at org.apache.lucene.search.CachingWrapperFilter.getDocIdSet(CachingWrapperFilter.java:203)
at org.apache.lucene.search.IndexSearcher.searchWithFilter(IndexSearcher.java:551)
at org.apache.lucene.search.IndexSearcher.search(IndexSearcher.java:532)
at org.apache.lucene.search.IndexSearcher.search(IndexSearcher.java:463)
at org.apache.lucene.search.IndexSearcher.search(IndexSearcher.java:433)
at org.apache.lucene.search.IndexSearcher.search(IndexSearcher.java:356)