[
https://issues.apache.org/jira/browse/HDFS-11597?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16030506#comment-16030506
]
Tsz Wo Nicholas Sze commented on HDFS-11597:
--------------------------------------------
> The question I was trying to ask here was – What happens if the SCM restarts
> with this change ? Does SCM remember this Ratis cluster information ? Or do
> we expect to relearn that info from heartbeats or something ?
This patch does not try to solve this problem. All the states regarding to
RatisManager currently is in memory but not persist to disk. As you mentioned,
do we want to persist the Ratis cluster states in SCM? Or just use heartbeats
to relearn them? I think we should persist the Ratis cluster states since they
correspond to the open containers (For closed containers, the Ratis cluster is
closed).
How does SCM current persist other states? We should do the same for
RatisManager.
> Ozone: Add Ratis management API
> -------------------------------
>
> Key: HDFS-11597
> URL: https://issues.apache.org/jira/browse/HDFS-11597
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Components: ozone
> Reporter: Tsz Wo Nicholas Sze
> Assignee: Tsz Wo Nicholas Sze
> Attachments: HDFS-11597-HDFS-7240.20170522.patch,
> HDFS-11597-HDFS-7240.20170523.patch, HDFS-11597-HDFS-7240.20170524.patch,
> HDFS-11597-HDFS-7240.20170528b.patch, HDFS-11597-HDFS-7240.20170528.patch,
> HDFS-11597-HDFS-7240.20170529.patch
>
>
> We need APIs to manage Ratis clusters for the following operations:
> - create cluster;
> - close cluster;
> - get members; and
> - update members.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
