[ https://issues.apache.org/jira/browse/HDFS-2339?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13106096#comment-13106096 ]
Uma Maheswara Rao G commented on HDFS-2339: ------------------------------------------- It looks to me that backup node is waiting when all volumes failed. {code} LOG.info("Waiting until the NameNode rolls its edit logs in order " + "to freeze the BackupNode namespace."); while (bnState == BNState.IN_SYNC) { Preconditions.checkState(stopApplyingEditsOnNextRoll, "If still in sync, we should still have the flag set to " + "freeze at next roll"); try { wait(); } catch (InterruptedException ie) { LOG.warn("Interrupted waiting for namespace to freeze", ie); throw new IOException(ie); } } {code} NameNode already removed the back up journals on failure. So, it is not giving any call to BNN. Since the backupnode streams already insync before , it is just waiting. Why do we need to run the process at this stage? > BackUpNode is not getting shutdown/recover when all volumes failed > ------------------------------------------------------------------ > > Key: HDFS-2339 > URL: https://issues.apache.org/jira/browse/HDFS-2339 > Project: Hadoop HDFS > Issue Type: Bug > Components: name-node > Affects Versions: 0.24.0 > Reporter: Uma Maheswara Rao G > Assignee: Uma Maheswara Rao G > > When all volumes failed at back up node side, it is waiting for ever. -- This message is automatically generated by JIRA. For more information on JIRA, see: http://www.atlassian.com/software/jira