hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Vinayakumar B (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HDFS-8929) Add a metric to expose the timestamp of the last journal
Date Tue, 08 Sep 2015 12:39:46 GMT

    [ https://issues.apache.org/jira/browse/HDFS-8929?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14734747#comment-14734747

Vinayakumar B commented on HDFS-8929:

Changes looks great.

Just one more improvement in Test.

Timestamp update can be verified after sending every edits in TestJournalNode#testJournal().

+1 once addressed.

> Add a metric to expose the timestamp of the last journal
> --------------------------------------------------------
>                 Key: HDFS-8929
>                 URL: https://issues.apache.org/jira/browse/HDFS-8929
>             Project: Hadoop HDFS
>          Issue Type: New Feature
>          Components: journal-node
>            Reporter: Akira AJISAKA
>            Assignee: Surendra Singh Lilhore
>         Attachments: HDFS-8929-001.patch, HDFS-8929-002.patch, HDFS-8929-003.patch
> If there are three JNs and only one JN is failing to journal, we can detect it by monitoring
the difference of the last written transaction id among JNs from NN WebUI or JN metrics. However,
it's difficult to define the threshold to alert because the increase rate of the number of
transaction depends on how busy the cluster is. Therefore I'd like to propose a metric to
expose the timestamp of the last journal. That way we can easily alert if a JN is failing
to journal for some fixed period.

This message was sent by Atlassian JIRA

View raw message