[ https://issues.apache.org/jira/browse/HBASE-20789?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16524848#comment-16524848 ]
Duo Zhang commented on HBASE-20789: ----------------------------------- I'm not familiar with HBASE-20447 but I'm a bit nervous that caching an already cached block. It may introduce memory leak as we will lost the pointer to the old block? And seems the original code after HBASE-13301 still have possible memory leak, as things could happen like this, we checked backing map, it is not there, and then a Writer thread write into the backing map, and remove it from the ramCache, and then we check ramCache, it is not there either, then we put it into the writer queue and cause recaching an existent block... > TestBucketCache#testCacheBlockNextBlockMetadataMissing is flaky > --------------------------------------------------------------- > > Key: HBASE-20789 > URL: https://issues.apache.org/jira/browse/HBASE-20789 > Project: HBase > Issue Type: Bug > Reporter: Zheng Hu > Assignee: Zheng Hu > Priority: Major > Attachments: bucket-33718.out > > > The UT failed frequently in our internal branch-2... Will dig into the UT. -- This message was sent by Atlassian JIRA (v7.6.3#76005)