[
https://issues.apache.org/jira/browse/YARN-949?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Jason Lowe reassigned YARN-949:
-------------------------------
Assignee: Kihwal Lee
The change is OK in the sense that it makes it more robust to errors occurring
during log uploading, but we're masking an NPE here which seems bad.
I think we should also fix the cause of the NPE itself, which is in
LogValue.write. This code in 0.23 should have a null check for {{in}} before
trying to call close:
{code}
// Write the log itself
FileInputStream in = null;
try {
in = new FileInputStream(logFile);
byte[] buf = new byte[65535];
int len = 0;
while ((len = in.read(buf)) != -1) {
out.write(buf, 0, len);
}
} finally {
in.close();
}
{code}
Looks like this was already fixed in trunk and branch-2 by YARN-578.
> Failed log aggregation can leave a file open.
> ---------------------------------------------
>
> Key: YARN-949
> URL: https://issues.apache.org/jira/browse/YARN-949
> Project: Hadoop YARN
> Issue Type: Bug
> Components: nodemanager
> Affects Versions: 2.1.0-beta, 0.23.9
> Reporter: Kihwal Lee
> Assignee: Kihwal Lee
> Attachments: YARN-949.patch
>
>
> If log aggregation fails on a node manager, the output file in hdfs can be
> left open.
--
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
