[
https://issues.apache.org/jira/browse/HDFS-14401?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16833623#comment-16833623
]
Feilong He commented on HDFS-14401:
-----------------------------------
HDFS-14401.008.patch has been uploaded with fixing the comments from [~rakeshr]
& [~anoop.hbase]. [~anoop.hbase] pointed out offline that there is potential
risk to drop user's data in pmem cache clean up logic. In this new patch, a
secondary dir will be created with conventional name. Thus the cache/delete
operations will be restricted under this dir, which can avoid user-data lost
risk in most cases. I also added cache cleanup logic into #Datanode.shutdown
when DN shuts down gracefully. Here, gracefully means DN's #shutdown is called
in some cases. But for 'stop-dfs.sh' command, DN's #shutdown is not called in
the current HDFS impl. Changing this behavior may lead to some unknown risks
and it needs more investigations.
> Refine the implementation for HDFS cache on SCM
> -----------------------------------------------
>
> Key: HDFS-14401
> URL: https://issues.apache.org/jira/browse/HDFS-14401
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Components: caching, datanode
> Reporter: Feilong He
> Assignee: Feilong He
> Priority: Major
> Attachments: HDFS-14401.000.patch, HDFS-14401.001.patch,
> HDFS-14401.002.patch, HDFS-14401.003.patch, HDFS-14401.004.patch,
> HDFS-14401.005.patch, HDFS-14401.006.patch, HDFS-14401.007.patch,
> HDFS-14401.008.patch
>
>
> In this Jira, we will refine the implementation for HDFS cache on SCM, such
> as: 1) Handle full pmem volume in VolumeManager; 2) Refine pmem volume
> selection impl; 3) Clean up MapppableBlockLoader interface; etc.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
