[
https://issues.apache.org/jira/browse/BLUR-105?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Aaron McCurry closed BLUR-105.
------------------------------
Resolution: Fixed
Fix Version/s: (was: 0.3.0)
0.2.4
This issue has been resolved for some amount time.
> Index Import rollback
> ---------------------
>
> Key: BLUR-105
> URL: https://issues.apache.org/jira/browse/BLUR-105
> Project: Apache Blur
> Issue Type: Improvement
> Components: Blur
> Affects Versions: 0.2.0, 0.3.0
> Reporter: Aaron McCurry
> Priority: Blocker
> Fix For: 0.2.4
>
>
> During the index import (after a map reduce job) if there is an exception
> thrown or a server shutdown the imported (and committed data from the
> MapReduce job) could be lost. This is because instead of copying the
> segments from the import directories into the main index. The index importer
> merely renames them and the reason for this is to improve performance.
> I suggest that we create a import logger to log what files have moved where
> so that in the event that the import fails, the files that have been moved
> into the main index and can be saved. I may need to create a diagram to
> illustrate the potential problem.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
