[
https://issues.apache.org/jira/browse/HDFS-11251?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Manoj Govindassamy updated HDFS-11251:
--------------------------------------
Attachment: HDFS-11251.01.patch
Attaching v01 patch to address the following.
1. Made {{Storage#storageDirs}} a CopyOnWriteArrayList instead of a normal
ArrayList so that concurrent iterators and writers don't end up in
{{ConcurrentModificationException}}. In this case, substitution with
CopyOnWriteArrayList is ok and should not cause any peformance degradation as
Volume add and removal are not frequent operations.
2. Tests updated to add more volumes during write and to expose the race
condition. On a private branched I recreated the original CME problem with this
test along with some delays introduced in {{Storage#containsStorageDir}}
This patch will throw a CheckStyle issue on {{Storage#storageDirs}} not being
private. But, this checkstyle issue existed even before the patch and touching
the same line again will cause the checkstyle to show up again. Will file a new
jira to cleanup the specifier for {{Storage#storageDirs}} and its access
patterns in the derived classes.
[~eddyxu], [~xiaochen], [~linyiqun], Can you please review the patch ?
> ConcurrentModificationException during DataNode#refreshVolumes
> --------------------------------------------------------------
>
> Key: HDFS-11251
> URL: https://issues.apache.org/jira/browse/HDFS-11251
> Project: Hadoop HDFS
> Issue Type: Bug
> Affects Versions: 3.0.0-alpha2
> Reporter: Jason Lowe
> Assignee: Manoj Govindassamy
> Attachments: HDFS-11251.01.patch
>
>
> The testAddVolumesDuringWrite case failed with a ReconfigurationException
> which appears to have been caused by a ConcurrentModificationException.
> Stacktrace details to follow.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
