hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Aaron T. Myers (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (HDFS-2100) Improve TestStorageRestore
Date Tue, 21 Jun 2011 22:35:47 GMT

     [ https://issues.apache.org/jira/browse/HDFS-2100?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel

Aaron T. Myers updated HDFS-2100:

    Attachment: hdfs-2100.1.patch

Updated patch addressing Todd's comments.

Seems like you got distracted halfway through writing this comment:
// The created file should still exist in the

Quite right. Changed to:

// The created file should still exist after the restart.

You might consider using ImmutableSet.of(path1) and ImmutableSet.of(path2, path3) from guava
- it's a bit easier to read IMO but no biggy

That's a great tip. I hated writing that code just to create a briefly-lived set.

Can you add a javadoc to explain what bug this is testing for and what the simulation is?
Seems funny that the test case is called testMultipleSecondasryCheckpoint but it only instantiates
one 2NN.

Added the following:

   * Test to simulate interleaved checkpointing by 2 2NNs after a storage
   * directory has been taken offline. The first will cause the directory to
   * come back online, but it won't have any valid contents. The second 2NN will
   * then try to perform a checkpoint. The NN should not serve up the image or
   * edits from the restored (empty) dir.

> Improve TestStorageRestore
> --------------------------
>                 Key: HDFS-2100
>                 URL: https://issues.apache.org/jira/browse/HDFS-2100
>             Project: Hadoop HDFS
>          Issue Type: Test
>          Components: test
>    Affects Versions: 0.23.0
>            Reporter: Aaron T. Myers
>            Assignee: Aaron T. Myers
>            Priority: Minor
>             Fix For: 0.23.0
>         Attachments: hdfs-2100.0.patch, hdfs-2100.1.patch
> Though running multiple 2NNs isn't supported, accidentally doing so should not result
in HDFS metadata corruptions. We should add a test case to exercise this possibility when
name.dir.storage.restore is enabled, which is a particularly delicate code path.

This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira


View raw message