[ https://issues.apache.org/jira/browse/BEAM-2140?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16066759#comment-16066759
]
Aljoscha Krettek commented on BEAM-2140:
----------------------------------------
I think to unblock this we have to move the SplittableDoFn implementation (in Flink, I don't
know about Dataflow) to a custom implementation that doesn't use {{ProcessFn}} because stateful/timely
DoFns are not a perfect fit right now. I did this in this branch: https://github.com/aljoscha/beam/tree/fix-flink-splittable-dofn-squashed
It works because we're now using Flink's processing-time timer facility directly, which doesn't
drop timers if they're late. This still has the issue of processing-time timers being simply
dropped when the pipeline shuts down (which happens when the bounded source shuts down).
About the more thorny issues regarding watermarks and splittable DoFn I feel we have to at
least bring the discussion to the ML and somehow design us out of the situation. 😃 What
do you think?
> Fix SplittableDoFn ValidatesRunner tests in FlinkRunner
> -------------------------------------------------------
>
> Key: BEAM-2140
> URL: https://issues.apache.org/jira/browse/BEAM-2140
> Project: Beam
> Issue Type: Bug
> Components: runner-flink
> Reporter: Aljoscha Krettek
> Assignee: Aljoscha Krettek
>
> As discovered as part of BEAM-1763, there is a failing SDF test. We disabled the tests
to unblock the open PR for BEAM-1763.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
|