[
https://issues.apache.org/jira/browse/BEAM-5036?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16597108#comment-16597108
]
Jozef Vilcek edited comment on BEAM-5036 at 8/30/18 6:49 AM:
-------------------------------------------------------------
[~timrobertson100] I guess if rename() API is kept defensive (throw exception
when target exists) that component invoking it must decide it is OK to
overwrite target or not. In case of `WriteOperation.moveToOutput()`, it clearly
is instructed to reconstruct the target from source data and I believe it
should obey. Job should be failed soon, during the launch if this is not
desired (e.g. target dir must be empty).
Re-run batch job is not the only case. In streaming, in case of failure ( or
upgrade ), job can be started from checkpoint. Depending on checkpoint time and
runner guarantees, some operations can be re-run and output files recreated
again. Operator must either delete target before `rename()` or use
`rename(overwrite = true)` if such choice would exists in API (delete/overwrite
off course in case source is present an there are actually data to be moved)
was (Author: jozovilcek):
[~timrobertson100] I guess if rename() API is kept defensive (throw exception
when target exists) that component invoking it must decide it is OK to
overwrite target or not. In case of `WriteOperation.moveToOutput()`, it clearly
is instructed to reconstruct the target from source data and I believe it
should obey. Job should be failed soon, during the launch if this is not
desired (e.g. target dir must be empty).
Re-run batch job is not the only case. In streaming, in case of failure ( or
upgrade ), job can be started from checkpoint. Depending on checkpoint time and
runner guarantees, some operations can be re-run and output files recreated
again. Operator must either delete target before `rename()` or use
`rename(overwrite = true)` if such choice would exists in API.
> Optimize FileBasedSink's WriteOperation.moveToOutput()
> ------------------------------------------------------
>
> Key: BEAM-5036
> URL: https://issues.apache.org/jira/browse/BEAM-5036
> Project: Beam
> Issue Type: Improvement
> Components: io-java-files
> Affects Versions: 2.5.0
> Reporter: Jozef Vilcek
> Assignee: Tim Robertson
> Priority: Major
> Time Spent: 1.5h
> Remaining Estimate: 0h
>
> moveToOutput() methods in FileBasedSink.WriteOperation implements move by
> copy+delete. It would be better to use a rename() which can be much more
> effective for some filesystems.
> Filesystem must support cross-directory rename. BEAM-4861 is related to this
> for the case of HDFS filesystem.
> Feature was discussed here:
> http://mail-archives.apache.org/mod_mbox/beam-dev/201807.mbox/%3CCAF9t7_4Mp54pQ+vRrJrBh9Vx0=uaknupzd_qdh_qdm9vxll...@mail.gmail.com%3E
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
