[ https://issues.apache.org/jira/browse/HDFS-2240?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Todd Lipcon updated HDFS-2240:
------------------------------
Attachment: hdfs-2240-2.txt
Even with the first patch, this didn't eliminate the cycles. Attaching a patch which does.
I also removed the now-unused "dfsc" argument from LeaseRenewer.get.
I agree this deadlock would not happen in practice, but having clean jcarder runs makes it
easier to detect when we do introduce real deadlocks.
> Possible deadlock between LeaseRenewer and its factory
> ------------------------------------------------------
>
> Key: HDFS-2240
> URL: https://issues.apache.org/jira/browse/HDFS-2240
> Project: Hadoop HDFS
> Issue Type: Bug
> Components: name-node
> Affects Versions: 0.23.0
> Reporter: Todd Lipcon
> Assignee: Tsz Wo (Nicholas), SZE
> Priority: Critical
> Fix For: 0.23.0
>
> Attachments: h2240_20110808.patch, hdfs-2240-2.txt, lease-renewer-cycle.png
>
>
> Lock cycle detected by jcarder
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
|