[ https://issues.apache.org/jira/browse/HDFS-9232?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14955661#comment-14955661 ]
Walter Su commented on HDFS-9232: --------------------------------- yes.HDFS-8344 aims to close the file if the replia is missing for a long time. It forces close the file. This JIRA, if it's unable to complete block/ close the file, I delay the recovery and wait more replias to be reported. > Shouldn't start block recovery if block has no enough replicas > -------------------------------------------------------------- > > Key: HDFS-9232 > URL: https://issues.apache.org/jira/browse/HDFS-9232 > Project: Hadoop HDFS > Issue Type: Bug > Reporter: Walter Su > Assignee: Walter Su > Attachments: HDFS-9232.01.patch > > > from HDFS-8406: > {quote} > Before primary DN calls commitBlockSynchronization, it synchronized 2 RBW > replicas, and make them finalized. Then primary DN calls > commitBlockSynchronization, to complete the lastBlock and close the file. The > question is, your dfs.namenode.replication.min is 3, the last block can't be > completed. NameNode shouldn't issue blockRecovery in the first place because > lastBlock can't be completed anyway. > {quote} -- This message was sent by Atlassian JIRA (v6.3.4#6332)