hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Mukul Kumar Singh (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HDFS-12329) Ozone: Ratis: Readonly calls in XceiverClientRatis should use sendReadOnly
Date Thu, 14 Sep 2017 07:30:01 GMT

    [ https://issues.apache.org/jira/browse/HDFS-12329?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16165888#comment-16165888

Mukul Kumar Singh commented on HDFS-12329:

Thanks for the review [~cheersyang],

1. In MiniOzoneCluster the patch added a method waitDatanodeOzoneReady to check if a datanode
is in RUNNING state. I am wondering why this cannot be done via waitOzoneReady. Because SCMNodeManager
should manage the states of all DNs, if they are rebooted, SCM should be aware and keep number
of healthy nodes updated.
bq. waitOzoneReady works on heartbeat mechanism, in this case the datanode restarted and came
back up before SCM could receive a heartbeat, hence waitOzoneReady is not sufficient in this

2. As you mentioned in HDFS-12354 this patch fixed the UT failure in TestKeys, I am wondering
what was the problem and how that is fixed in this patch.
bq. I could not get any useful information from the Jenkins run, however because I had identified
this issue, I feel that both the issues might be related.

> Ozone: Ratis: Readonly calls in XceiverClientRatis should use sendReadOnly
> --------------------------------------------------------------------------
>                 Key: HDFS-12329
>                 URL: https://issues.apache.org/jira/browse/HDFS-12329
>             Project: Hadoop HDFS
>          Issue Type: Sub-task
>          Components: ozone
>    Affects Versions: HDFS-7240
>            Reporter: Mukul Kumar Singh
>            Assignee: Mukul Kumar Singh
>              Labels: OzonePostMerge
>             Fix For: HDFS-7240
>         Attachments: HDFS-12329-HDFS-7240.001.patch, HDFS-12329-HDFS-7240.002.patch
> Currently both write and readonly calls in Ratis use RaftClient.send which enqueues the
the request to the raft log and is processed later when the log entry is consumed.
> Readonly call can be optimized by using RaftClient.sendReadOnly which will directly query
the RaftServer for a particular request.
> This jira will be used to discuss this issue.

This message was sent by Atlassian JIRA

To unsubscribe, e-mail: hdfs-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-help@hadoop.apache.org

View raw message