[
https://issues.apache.org/jira/browse/HDFS-13655?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16592424#comment-16592424
]
Yiqun Lin commented on HDFS-13655:
----------------------------------
I'm also a little worried about the current JIRA numbers. I take a quick glance
of recent RBF JIRAs. There are not only error handling of the Router admin,
also have others following:
* HDFS-13845:The default MountTableResolver cannot get multi-destination path
for the DestinationOrder
* HDFS-13841: After Click on another Tab in Router Federation UI page it's not
redirecting to new tab
* HDFS-13835: RBF: Unable to add files after changing the order
* HDFS-13852: RBF: The DN_REPORT_TIME_OUT and DN_REPORT_CACHE_EXPIRE should be
configured in RBFConfigKeys.
* HDFS-13802: RBF: Remove FSCK from Router Web UI, because fsck is not
supported currently
For the critical issues, we can directly merge to trunk. For most of others
(not critical issues), I prefer to merge these JIRAs in the same branch and
make the branch stabilised.
So the new branch will focus on:
* ClientProtocol APIs implementation to RBF
* Router admin error handling improvments
* Other improvements(e.g. setting configurable, normal bug fixes)
One thing I'm sure, it's not appropriate to directly merge these all to trunk
now.
> RBF: Add missing ClientProtocol APIs to RBF
> -------------------------------------------
>
> Key: HDFS-13655
> URL: https://issues.apache.org/jira/browse/HDFS-13655
> Project: Hadoop HDFS
> Issue Type: Improvement
> Reporter: Xiao Chen
> Priority: Major
>
> As
> [discussed|https://issues.apache.org/jira/browse/HDFS-12858?focusedCommentId=16500975&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-16500975|#comment-16500975]
> with [~elgoiri], there are some HDFS methods that does not take path as a
> parameter. We should support these to work with federation.
> The ones missing are:
> * Snapshots
> * Storage policies
> * Encryption zones
> * Cache pools
> One way to reasonably have them to work with federation is to 'list' each
> nameservice and concat the results. This can be done pretty much the same as
> {{refreshNodes()}} and it would be a matter of querying all the subclusters
> and aggregate the output (e.g., {{getDatanodeReport()}}.)
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
