[
https://issues.apache.org/jira/browse/HDFS-5431?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13846732#comment-13846732
]
Andrew Wang commented on HDFS-5431:
-----------------------------------
Thanks Colin, I'll get to all the things you mentioned along with rebasing for
the HDFS-2832 changes. One note though:
On changing the refresh interval, I can see monitoring and apps polling
listPools/listDirectives pretty frequently. If the expected rate of external
triggers of CRM#refresh is higher than our own refresh interval, we might as
well just do it asynchronously at that expected rate rather than doing it
synchronously on every RPC. I know Cloudera Manager typically does these things
on a 1 minute tick, which is why I chose this value. A faster refresh interval
will also make caching more reactive, which is also nice, and we might be able
to get rid of the retry logic for resending caching commands too.
Another thing is the delay for updating the {{*Cached}} stats. It takes one
refresh to schedule caching work, and another refresh to update the stats.
Doing this incrementally means mapping blocks -> pools or directives, which
probably isn't going to happen.
> support cachepool-based limit management in path-based caching
> --------------------------------------------------------------
>
> Key: HDFS-5431
> URL: https://issues.apache.org/jira/browse/HDFS-5431
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Components: datanode, namenode
> Affects Versions: 3.0.0
> Reporter: Colin Patrick McCabe
> Assignee: Colin Patrick McCabe
> Attachments: hdfs-5431-1.patch, hdfs-5431-2.patch, hdfs-5431-3.patch
>
>
> We should support cachepool-based quota management in path-based caching.
--
This message was sent by Atlassian JIRA
(v6.1.4#6159)
