[ https://issues.apache.org/jira/browse/HDFS-14186?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16743281#comment-16743281 ]
Íñigo Goiri commented on HDFS-14186: ------------------------------------ {quote} As description above, if service port is overrun and some DataNode does not register successfully, lifeline could not fix it, FYI. Please correct me if there are something wrong. {quote} In tried to go through the registration part but it's not that easy to follow; I would need a proper pass. In any case, I think that if the DN is not registered, it cannot be marked as DEAD. I think, for a DN to show up in the NN, it needs to be fully registered. If that's the case, the lifeline will keep it as ALIVE. The issue might be the lock in {{HeartbeatManager#updateLifeline()}}. Could you post where the NN is spending most of the time? Ideally a stack trace of the thread that is holding the other requests. > blockreport storm slow down namenode restart seriously in large cluster > ----------------------------------------------------------------------- > > Key: HDFS-14186 > URL: https://issues.apache.org/jira/browse/HDFS-14186 > Project: Hadoop HDFS > Issue Type: Improvement > Components: namenode > Reporter: He Xiaoqiao > Assignee: He Xiaoqiao > Priority: Major > Attachments: HDFS-14186.001.patch > > > In the current implementation, the datanode sends blockreport immediately > after register to namenode successfully when restart, and the blockreport > storm will make namenode high load to process them. One result is some > received RPC have to skip because queue time is timeout. If some datanodes' > heartbeat RPC are continually skipped for long times (default is > heartbeatExpireInterval=630s) it will be set DEAD, then datanode has to > re-register and send blockreport again, aggravate blockreport storm and trap > in a vicious circle, and slow down (more than one hour and even more) > namenode startup seriously in a large (several thousands of datanodes) and > busy cluster especially. Although there are many work to optimize namenode > startup, the issue still exists. > I propose to postpone dead datanode check when namenode have finished startup. > Any comments and suggestions are welcome. -- This message was sent by Atlassian JIRA (v7.6.3#76005) --------------------------------------------------------------------- To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org