[
https://issues.apache.org/jira/browse/HUDI-945?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17294411#comment-17294411
]
wangmeng commented on HUDI-945:
-------------------------------
The method(clear) of DiskBasedMap.
Can we go here to close the stream and delete the file？
????
public void clear() {
valueMetadataMap.clear();
// Do not delete file-handles & file as there is no way to do it without
synchronizing get/put(and
// reducing concurrency). Instead, just clear the pointer map. The file will
be removed on exit.
}
> Cleanup spillable map files eagerly as part of close
> ----------------------------------------------------
>
> Key: HUDI-945
> URL: https://issues.apache.org/jira/browse/HUDI-945
> Project: Apache Hudi
> Issue Type: Bug
> Components: Writer Core
> Reporter: Balaji Varadarajan
> Assignee: Sreeram Ramji
> Priority: Major
> Labels: pull-request-available
> Fix For: 0.8.0
>
>
> Currently, files used by external spillable map are deleted on exits. For
> spark-streaming/deltastreamer continuous-mode cases which runs several
> iterations, it is better to eagerly delete files on closing the handles using
> it.
> We need to eagerly delete the files on following cases:
> # HoodieMergeHandle
> # HoodieMergedLogRecordScanner
> # SpillableMapBasedFileSystemView
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
