[
https://issues.apache.org/jira/browse/HADOOP-15604?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16835612#comment-16835612
]
Steve Loughran commented on HADOOP-15604:
-----------------------------------------
S3Guard.addAncestors() tries to efficiently walk up the tree and only call
put() on entries which don't exist, so avoiding that excessive load.
But: {{metadataStore.put(newDirs)}} goes on to create all the ancestors in
{{innerPut(Collection<DDBPathMetadata> metas)}}. That is: it doesn't bother
looking for the parent entries, it just blindly tries to create them all. For
HADOOP-15183 I'm minimising this across move operations by passing a context
around for the {{move()}} calls, I think this same idea somehow needs to be
preserved here, but its a lot harder to join up given that its
S3AFileSystem.finishedWrite() where this stuff is done and the context is
pretty minimal.
> Bulk commits of S3A MPUs place needless excessive load on S3 & S3Guard
> ----------------------------------------------------------------------
>
> Key: HADOOP-15604
> URL: https://issues.apache.org/jira/browse/HADOOP-15604
> Project: Hadoop Common
> Issue Type: Sub-task
> Components: fs/s3
> Affects Versions: 3.1.0
> Reporter: Gabor Bota
> Assignee: Steve Loughran
> Priority: Major
>
> When there are ~50 files being committed; each in their own thread from the
> commit pool; probably the DDB repo is being overloaded just from one single
> process doing task commit. We should be backing off more, especially given
> that failing on a write could potentially leave the store inconsistent with
> the FS (renames, etc)
> It would be nice to have some tests to prove that the I/O thresholds are the
> reason for unprocessed items in DynamoDB metadata store
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
