hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Ted Yu (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-4161) Incorrect use of listStatus() in HBase region initialization.
Date Fri, 05 Aug 2011 13:09:28 GMT

    [ https://issues.apache.org/jira/browse/HBASE-4161?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13079955#comment-13079955
] 

Ted Yu commented on HBASE-4161:
-------------------------------

Thanks for the update.

Looking at processDeleteOnExit in FileSystem.java:
{code}
        catch (IOException e) {
          LOG.info("Ignoring failure to deleteOnExit for path " + path);
        }
{code}
exists() throws IOException.
I wonder what other IOException may come out of exists(). Should we catch such exception ?

> Incorrect use of listStatus() in HBase region initialization.
> -------------------------------------------------------------
>
>                 Key: HBASE-4161
>                 URL: https://issues.apache.org/jira/browse/HBASE-4161
>             Project: HBase
>          Issue Type: Bug
>          Components: regionserver
>            Reporter: Pritam Damania
>         Attachments: 0001-Fix-FileNotFoundException-in-HLog.java.patch
>
>
> While opening a region, the HBase regionserver tries to list all the children in a "recovered.edits"
directory. This directory may not exist and depending on the version of HDFS listStatus()
might return null or an exception. If it does throw an exception the entire process of opening
the region is aborted, just because the recovered.edits directory is not present.

--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

Mime
View raw message