[
https://issues.apache.org/jira/browse/HBASE-23353?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Bo Cui updated HBASE-23353:
---------------------------
Description:
in PartitionedMobCompactor#bulkloadRefFile
If region num is changed from 1 to 20
bulkload will fail and we will get an exception (because the bulkload
maxRetries is 10，hbase.bulkload.retries.number)
!image-2019-11-30-10-30-33-243.png!
and then delete bulkload path and new MobFile,
old mob file is not deleted.
but some hfiles have been successfully loaded through bulkload and not deleted.
!image-2019-11-30-10-40-32-837.png!
And then,,,, get rowkey,,,,throw filenotfound
solution : write multiple hfiles and reduce split hfiles. because the data of
scanner is ordered，we just need to open a hfile at the same time, when the
current data dose not belong to the current hfile, we can close hfile and
create new hfile.
!image-2019-11-30-10-33-00-867.png!
was:
in PartitionedMobCompactor#bulkloadRefFile
If region num is changed from 1 to 20
bulkload will fail and we will get an exception
!image-2019-11-30-10-30-33-243.png!
and then delete bulkload path and new MobFile,
old mob file is not deleted.
but some hfiles have been successfully loaded through bulkload and not deleted.
!image-2019-11-30-10-40-32-837.png!
And then,,,, get rowkey,,,,throw filenotfound
solution : write multiple hfiles and reduce split hfiles. because the data of
scanner is ordered，we just need to open a hfile at the same time, when the
current data dose not belong to the current hfile, we can close hfile and
create new hfile.
!image-2019-11-30-10-33-00-867.png!
> MOB throws FileNotFound
> -----------------------
>
> Key: HBASE-23353
> URL: https://issues.apache.org/jira/browse/HBASE-23353
> Project: HBase
> Issue Type: Bug
> Components: mob
> Affects Versions: 2.1.1, 2.2.2
> Reporter: Bo Cui
> Priority: Critical
> Attachments: image-2019-11-30-10-30-33-243.png,
> image-2019-11-30-10-33-00-867.png, image-2019-11-30-10-40-32-837.png
>
>
> in PartitionedMobCompactor#bulkloadRefFile
> If region num is changed from 1 to 20
> bulkload will fail and we will get an exception (because the bulkload
> maxRetries is 10，hbase.bulkload.retries.number)
> !image-2019-11-30-10-30-33-243.png!
> and then delete bulkload path and new MobFile,
> old mob file is not deleted.
> but some hfiles have been successfully loaded through bulkload and not
> deleted.
> !image-2019-11-30-10-40-32-837.png!
> And then,,,, get rowkey,,,,throw filenotfound
> solution : write multiple hfiles and reduce split hfiles. because the data of
> scanner is ordered，we just need to open a hfile at the same time, when the
> current data dose not belong to the current hfile, we can close hfile and
> create new hfile.
> !image-2019-11-30-10-33-00-867.png!
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
