[ https://issues.apache.org/jira/browse/HBASE-7638?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13561823#comment-13561823 ]
Sergey Shelukhin commented on HBASE-7638: ----------------------------------------- The problems arises in multithreaded case (actually I only observed it in trunk, but I assumed this should also apply to 94). One thread gets A from cache and goes to A. Thread 2 does the same. Thread 1 errors out and updates the location from META to new correct location. Thread 2 errors out and removes it from cache. So it's not critical, just adds extra meta trips. In integration test where there are many threads and many errors it was relatively more prominent. I think it's made possible in no small part by using delayed callable, where we first build the request, including the destination server, then wait. > [0.94] region cache entry should only be removed on error if the error is > from the server currently in cache > ------------------------------------------------------------------------------------------------------------ > > Key: HBASE-7638 > URL: https://issues.apache.org/jira/browse/HBASE-7638 > Project: HBase > Issue Type: Bug > Affects Versions: 0.94.4 > Reporter: Sergey Shelukhin > Assignee: Sergey Shelukhin > Priority: Minor > Fix For: 0.94.5 > > Attachments: HBASE-7638-v0.patch > > > See HBASE-7268. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira