[
https://issues.apache.org/jira/browse/HADOOP-11924?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14559560#comment-14559560
]
Gera Shegalov commented on HADOOP-11924:
----------------------------------------
[~ozawa], are you going to work on 002? I think at the very least we should
change the log level when swallowing the exception. The exception itself should
also be included in the LOG statement:
{code}
LOG.info("Avoiding JDK-8047340 on BSD-based systems.", t);
{code}
> Tolerate JDK-8047340-related exceptions in Shell#isSetSidAvailable preventing
> class init
> ----------------------------------------------------------------------------------------
>
> Key: HADOOP-11924
> URL: https://issues.apache.org/jira/browse/HADOOP-11924
> Project: Hadoop Common
> Issue Type: Bug
> Affects Versions: 2.8.0
> Reporter: Gera Shegalov
> Assignee: Tsuyoshi Ozawa
> Attachments: HADOOP-11924.001.patch
>
>
> Address the root cause of HADOOP-11916 per
> https://issues.apache.org/jira/browse/HADOOP-11916?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14528009#comment-14528009
> {quote}
> JDK-8047340 explicitly calls out BSD-like systems, should not we just exclude
> those systems instead of enabling solely Linux?
> {code}
> Assume.assumeFalse("Avoiding JDK-8047340 on BSD-based systems", Shell.FREEBSD
> || Shell.MAC);
> {code}
> However, I don't think this is the right fix. Shell on BSD-like systems is
> broken with the TR locale. Shell class initialization happens only because
> StringUtils references Shell.WINDOWS.
> We can simply catch Throwable in Shell#isSetsidSupported instead of
> IOException. If we want to be pedantic we can rethrow
> {code}
> if (!(t instanceof IOException) && !(Shell.FREEBSD || Shell.MAC))
> {code}
> With such a change the test can run unchanged.
> {quote}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
