You are viewing a plain text version of this content. The canonical link for it is here.
Posted to hdfs-issues@hadoop.apache.org by "Kihwal Lee (JIRA)" <ji...@apache.org> on 2018/05/01 18:22:00 UTC
[jira] [Commented] (HDFS-13519) Namenode recovery does not replay
the inProgress edit log
[ https://issues.apache.org/jira/browse/HDFS-13519?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16459976#comment-16459976 ]
Kihwal Lee commented on HDFS-13519:
-----------------------------------
You can always manually issue "dfsadmin -rollEdits". Also you can set "dfs.namenode.edit.log.autoroll.multiplier.threshold" to make ActiveNN automatically roll edits, so that each segment doesn't get excessively big.
> Namenode recovery does not replay the inProgress edit log
> ---------------------------------------------------------
>
> Key: HDFS-13519
> URL: https://issues.apache.org/jira/browse/HDFS-13519
> Project: Hadoop HDFS
> Issue Type: Bug
> Reporter: Hanisha Koneru
> Priority: Major
>
> When we start Namenode with --recover flag set, the NN replays the transactions uptill the last finalized edit log. The inProgress edit log is not included in this. If for some reason checkpointing is not happening on the NN, then the inProgress edit log might contain a huge number of transactions.
> We should have an option to replay the inProgress edit logs also when doing a recovery on the Namenode.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-help@hadoop.apache.org