[
https://issues.apache.org/jira/browse/HDFS-8546?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Andrew Wang updated HDFS-8546:
------------------------------
Attachment: HDFS-8546.002.patch
Looking at it more, I realized that copying the dncp log over on upgrade is
unnecessary, since the O(1) BlockScanner rewrite doesn't use it. The dncp log
will still be in the previous directory if the DN is rolled back, but there's
no need to keep carrying it forward.
> Use try with resources in DataStorage and Storage
> -------------------------------------------------
>
> Key: HDFS-8546
> URL: https://issues.apache.org/jira/browse/HDFS-8546
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: datanode
> Affects Versions: 2.7.0
> Reporter: Andrew Wang
> Assignee: Andrew Wang
> Priority: Minor
> Attachments: HDFS-8546.001.patch, HDFS-8546.002.patch
>
>
> We have some old-style try/finally to close files in DataStorage and Storage,
> let's update them.
> Also a few small cleanups:
> * Actually check that tryLock returns a FileLock in isPreUpgradableLayout
> * Remove unused parameter from writeProperties
> * Add braces for one-line if statements per coding style
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
