[
https://issues.apache.org/jira/browse/HDFS-11941?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16042828#comment-16042828
]
Allen Wittenauer commented on HDFS-11941:
-----------------------------------------
metasave is absolutely critical when the on-disk copy of the fsimage file is
known bad but not enough for the NN to crash.
> Move dfsadmin triggerBlockReport and metaSave to debugadmin
> -----------------------------------------------------------
>
> Key: HDFS-11941
> URL: https://issues.apache.org/jira/browse/HDFS-11941
> Project: Hadoop HDFS
> Issue Type: Improvement
> Affects Versions: 2.9.0, 3.0.0-alpha4
> Reporter: Andrew Wang
>
> Filing a JIRA for discussion. While reviewing the [dfsadmin
> commands|https://hadoop.apache.org/docs/current/hadoop-project-dist/hadoop-hdfs/HDFSCommands.html#dfsadmin]
> some of them seem better suited for debugadmin:
> * triggerBlockReport: similar to recoverLease in debugadmin, you don't need
> this unless HDFS is in a bad state
> * metasave: dumps NN datastructures to a side file. Seems purely for
> debugging.
> DebugAdmin commands notably do not need to be compatible between releases and
> do not have Java API equivalents in HdfsAdmin.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
