Hernan Gelaf-Romer created HBASE-29146:
------------------------------------------
Summary: Incremental Backups Fail
Key: HBASE-29146
URL: https://issues.apache.org/jira/browse/HBASE-29146
Project: HBase
Issue Type: Bug
Components: backup&restore
Reporter: Hernan Gelaf-Romer
At the moment, if a bulkload file is archived as we're backing it up, or if
there are archive files that need to be loaded, we run into the following
exception
```
Unexpected Exception : org.apache.hadoop.mapred.FileAlreadyExistsException
```
The happens because we're running MR jobs to re-split these HFiles, and we
cannot run subsequent MR jobs against the same target bulk output directory
without first cleaning that directory up. We should make sure we properly clean
up the bulkload output directory in between runs of MapReduceHFileSplitterJob
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
