[ https://issues.apache.org/jira/browse/HBASE-18092?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Ashu Pachauri updated HBASE-18092: ---------------------------------- Attachment: HBASE-18092.branch-1.3.001.patch HBASE-18092.branch-1.001.patch HBASE-18092.master.002.patch Attaching another patch for master. I found a redundant piece of code related to the patch. The only difference between the previous and this patch is {Code} @@ -528,9 +542,7 @@ public class ReplicationSourceManager implements ReplicationListener { */ public void closeRecoveredQueue(ReplicationSourceInterface src) { LOG.info("Done with the recovered queue " + src.getPeerClusterZnode()); - if (src instanceof ReplicationSource) { - ((ReplicationSource) src).getSourceMetrics().clear(); - } + src.getSourceMetrics().clear(); {Code} Also attaching patches for branch-1 and branch-1.3 as they both have diverged a little bit. > Removing a peer does not properly clean up the ReplicationSourceManager state > and metrics > ----------------------------------------------------------------------------------------- > > Key: HBASE-18092 > URL: https://issues.apache.org/jira/browse/HBASE-18092 > Project: HBase > Issue Type: Bug > Components: Replication > Affects Versions: 2.0.0, 1.3.1 > Reporter: Ashu Pachauri > Assignee: Ashu Pachauri > Attachments: HBASE-18092.branch-1.001.patch, > HBASE-18092.branch-1.3.001.patch, HBASE-18092.master.001.patch, > HBASE-18092.master.002.patch > > > Removing a peer does not clean up the associated metrics and state from > walsById map in the ReplicationSourceManager. -- This message was sent by Atlassian JIRA (v6.3.15#6346)