hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Todd Lipcon (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (HDFS-3884) QJM: Journal format() should reset cached values
Date Wed, 05 Sep 2012 04:28:10 GMT

     [ https://issues.apache.org/jira/browse/HDFS-3884?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel

Todd Lipcon updated HDFS-3884:

    Attachment: hdfs-3884.txt

Attaching patch I'm about to commit. Had to slightly modify it due to changes in the order
the patches were staged vs committed. For it to compile, had to add the following getter which
was from the patch in HDFS-3870:

+  synchronized public long getLastWriterEpoch() throws IOException {
+    checkFormatted();
+    return lastWriterEpoch.get();
+  }

Since it's a trivial change, I'll commit based on the earlier +1s.
> QJM: Journal format() should reset cached values
> ------------------------------------------------
>                 Key: HDFS-3884
>                 URL: https://issues.apache.org/jira/browse/HDFS-3884
>             Project: Hadoop HDFS
>          Issue Type: Sub-task
>    Affects Versions: QuorumJournalManager (HDFS-3077)
>            Reporter: Todd Lipcon
>            Assignee: Todd Lipcon
>             Fix For: QuorumJournalManager (HDFS-3077)
>         Attachments: hdfs-3884.txt, hdfs-3884.txt, hdfs-3884.txt
> Simple bug in the JournalNode: it caches certain values (eg accepted epoch) in memory,
and the cached values aren't reset when the journal is formatted. So, after a format, further
calls to the same Journal will see the old value for accepted epoch, writer epoch, etc, preventing
the journal from being re-used until the JN is restarted.

This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

View raw message