[
https://issues.apache.org/jira/browse/BEAM-2680?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16899655#comment-16899655
]
Steve Niemitz commented on BEAM-2680:
-------------------------------------
I ran into this today. We have a long-running streaming pipeline that watches
a GCS bucket and processes files as they arrive. It eventually was no longer
able to make progress because the GrowthState (completed) was too big to fit
into memory.
> Improve scalability of the Watch transform
> ------------------------------------------
>
> Key: BEAM-2680
> URL: https://issues.apache.org/jira/browse/BEAM-2680
> Project: Beam
> Issue Type: Bug
> Components: sdk-java-core
> Reporter: Eugene Kirpichov
> Assignee: Eugene Kirpichov
> Priority: Major
>
> [https://github.com/apache/beam/pull/3565] introduces the Watch transform
> [http://s.apache.org/beam-watch-transform].
> The implementation leaves several scalability-related TODOs:
> 1) The state stores hashes and timestamps of outputs that have already been
> output and should be omitted from future polls. We could garbage-collect this
> state, e.g. dropping elements from "completed" and from addNewAsPending() if
> their timestamp is more than X behind the watermark.
> 2) When a poll returns a huge number of elements, we don't necessarily have
> to add all of them into state.pending - instead we could add only N oldest
> elements and ignore others, relying on future poll rounds to provide them, in
> order to avoid blowing up the state. Combined with garbage collection of
> GrowthState.completed, this would make the transform scalable to very large
> poll results.
--
This message was sent by Atlassian JIRA
(v7.6.14#76016)
