hadoop-yarn-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Sangjin Lee (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (YARN-3949) ensure timely flush of timeline writes
Date Fri, 24 Jul 2015 19:37:06 GMT

    [ https://issues.apache.org/jira/browse/YARN-3949?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14640974#comment-14640974

Sangjin Lee commented on YARN-3949:

For the hbase writer, {{flush()}} and {{close()}} (or any other method for that matter) are
synchronized. So they will not execute concurrently.

I think the only difference between {{shutdown()}} and {{shutdownNow()}} in this case is whether
we would want to interrupt the running {{flush()}} and go straight to {{close()}} (for {{shutdownNow()}}),
or let {{flush()}} finish and go to {{close()}} (for {{shutdown()}}). Again, that's only for

Since {{shutdown()}} is more graceful and will let {{flush()}} finish, I'm bit more inclined
to do that now. I'd still need to introduce a call to {{awaitTermination()}} to be correct
in all cases then although the hbase writer doesn't really need it. Let me see what I can
do, along with those comments for the configuration.

I'm stepping out for now, but will post a new patch by EOD.

> ensure timely flush of timeline writes
> --------------------------------------
>                 Key: YARN-3949
>                 URL: https://issues.apache.org/jira/browse/YARN-3949
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>          Components: timelineserver
>    Affects Versions: YARN-2928
>            Reporter: Sangjin Lee
>            Assignee: Sangjin Lee
>         Attachments: YARN-3949-YARN-2928.001.patch, YARN-3949-YARN-2928.002.patch, YARN-3949-YARN-2928.002.patch
> Currently flushing of timeline writes is not really handled. For example, {{HBaseTimelineWriterImpl}}
relies on HBase's {{BufferedMutator}} to batch and write puts asynchronously. However, {{BufferedMutator}}
may not flush them to HBase unless the internal buffer fills up.
> We do need a flush functionality first to ensure that data are written in a reasonably
timely manner, and to be able to ensure some critical writes are done synchronously (e.g.
key lifecycle events).

This message was sent by Atlassian JIRA

View raw message