You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@hbase.apache.org by "stack (JIRA)" <ji...@apache.org> on 2008/05/28 22:58:45 UTC
[jira] Created: (HBASE-648) If mapfile index is empty, run repair
If mapfile index is empty, run repair
-------------------------------------
Key: HBASE-648
URL: https://issues.apache.org/jira/browse/HBASE-648
Project: Hadoop HBase
Issue Type: Improvement
Reporter: stack
Our cluster lost data. Made for some interesting scenarios in hbase. One such was empty index files (Would get an EOFException when we tried to scan or open region). HBASE-646 added checking of data, info, and index files. Tried to run repair of indexes but was doing it wrong place and wasn't first removing the broken index so it'd fail.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Resolved: (HBASE-648) If mapfile index is empty, run repair
Posted by "stack (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-648?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
stack resolved HBASE-648.
-------------------------
Resolution: Fixed
Committed to trunk and branch.
Of note, the repair uses a hardcoded interval of 128 rather than our configurable interval w/ a default of 32.
> If mapfile index is empty, run repair
> -------------------------------------
>
> Key: HBASE-648
> URL: https://issues.apache.org/jira/browse/HBASE-648
> Project: Hadoop HBase
> Issue Type: Improvement
> Reporter: stack
> Assignee: stack
> Attachments: fix.patch
>
>
> Our cluster lost data. Made for some interesting scenarios in hbase. One such was empty index files (Would get an EOFException when we tried to scan or open region). HBASE-646 added checking of data, info, and index files. Tried to run repair of indexes but was doing it wrong place and wasn't first removing the broken index so it'd fail.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (HBASE-648) If mapfile index is empty, run repair
Posted by "stack (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-648?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12600624#action_12600624 ]
stack commented on HBASE-648:
-----------------------------
That should be, that the MapFile.fix function has a hardcoded interval of 128.
> If mapfile index is empty, run repair
> -------------------------------------
>
> Key: HBASE-648
> URL: https://issues.apache.org/jira/browse/HBASE-648
> Project: Hadoop HBase
> Issue Type: Improvement
> Reporter: stack
> Assignee: stack
> Attachments: fix.patch
>
>
> Our cluster lost data. Made for some interesting scenarios in hbase. One such was empty index files (Would get an EOFException when we tried to scan or open region). HBASE-646 added checking of data, info, and index files. Tried to run repair of indexes but was doing it wrong place and wasn't first removing the broken index so it'd fail.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Assigned: (HBASE-648) If mapfile index is empty, run repair
Posted by "stack (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-648?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
stack reassigned HBASE-648:
---------------------------
Assignee: stack
> If mapfile index is empty, run repair
> -------------------------------------
>
> Key: HBASE-648
> URL: https://issues.apache.org/jira/browse/HBASE-648
> Project: Hadoop HBase
> Issue Type: Improvement
> Reporter: stack
> Assignee: stack
> Attachments: fix.patch
>
>
> Our cluster lost data. Made for some interesting scenarios in hbase. One such was empty index files (Would get an EOFException when we tried to scan or open region). HBASE-646 added checking of data, info, and index files. Tried to run repair of indexes but was doing it wrong place and wasn't first removing the broken index so it'd fail.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (HBASE-648) If mapfile index is empty, run repair
Posted by "stack (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-648?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
stack updated HBASE-648:
------------------------
Attachment: fix.patch
{code}
M src/java/org/apache/hadoop/hbase/HStoreFile.java
Genericized the static defines of classes used in HbaseMapFile.
Fixed javadoc.
M src/java/org/apache/hadoop/hbase/HStore.java
Don't catch the EOFException opening readers. Instead, run code
earlier that checks for empty index and run repairs there.
(isEmptyIndexFile, isEmptyFile): Added.
{code}
> If mapfile index is empty, run repair
> -------------------------------------
>
> Key: HBASE-648
> URL: https://issues.apache.org/jira/browse/HBASE-648
> Project: Hadoop HBase
> Issue Type: Improvement
> Reporter: stack
> Attachments: fix.patch
>
>
> Our cluster lost data. Made for some interesting scenarios in hbase. One such was empty index files (Would get an EOFException when we tried to scan or open region). HBASE-646 added checking of data, info, and index files. Tried to run repair of indexes but was doing it wrong place and wasn't first removing the broken index so it'd fail.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (HBASE-648) If mapfile index is empty, run repair
Posted by "Jim Kellerman (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-648?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Jim Kellerman updated HBASE-648:
--------------------------------
Fix Version/s: 0.2.0
> If mapfile index is empty, run repair
> -------------------------------------
>
> Key: HBASE-648
> URL: https://issues.apache.org/jira/browse/HBASE-648
> Project: Hadoop HBase
> Issue Type: Improvement
> Reporter: stack
> Assignee: stack
> Fix For: 0.2.0
>
> Attachments: fix.patch
>
>
> Our cluster lost data. Made for some interesting scenarios in hbase. One such was empty index files (Would get an EOFException when we tried to scan or open region). HBASE-646 added checking of data, info, and index files. Tried to run repair of indexes but was doing it wrong place and wasn't first removing the broken index so it'd fail.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.