[
https://issues.apache.org/jira/browse/BEAM-7520?focusedWorklogId=326071&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-326071
]
ASF GitHub Bot logged work on BEAM-7520:
----------------------------------------
Author: ASF GitHub Bot
Created on: 10/Oct/19 02:40
Start Date: 10/Oct/19 02:40
Worklog Time Spent: 10m
Work Description: kennknowles commented on pull request #9190:
[BEAM-7520] Fix timer firing order in DirectRunner
URL: https://github.com/apache/beam/pull/9190#discussion_r333306703
##########
File path:
runners/direct-java/src/main/java/org/apache/beam/runners/direct/StatefulParDoEvaluatorFactory.java
##########
@@ -254,17 +268,23 @@ public void
processElement(WindowedValue<KeyedWorkItem<K, KV<K, InputT>>> gbkRes
WindowNamespace<?> windowNamespace = (WindowNamespace)
timer.getNamespace();
BoundedWindow timerWindow = windowNamespace.getWindow();
delegateEvaluator.onTimer(timer, timerWindow);
+ if (timerInternals.containsUpdateForTimeBefore(currentInputWatermark))
{
Review comment:
So if I understand this correctly, the `PriorityQueue` causes the timers in
the incoming bundle to be fired in timestamp order. That's good. And then if
any timer is set that would already be eligible for firing, it ends processing
and pushes the rest back. I think this is not optimal, because you could add
timer updates to the priority queue and keep firing any eligible timer. The
bundle only needs to end when there are no more eligible timers. But on the
other hand, I do think it fixes the problem, so happy to merge it.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
Issue Time Tracking
-------------------
Worklog Id: (was: 326071)
Time Spent: 10h 20m (was: 10h 10m)
> DirectRunner timers are not strictly time ordered
> -------------------------------------------------
>
> Key: BEAM-7520
> URL: https://issues.apache.org/jira/browse/BEAM-7520
> Project: Beam
> Issue Type: Bug
> Components: runner-direct
> Affects Versions: 2.13.0
> Reporter: Jan Lukavský
> Assignee: Jan Lukavský
> Priority: Major
> Time Spent: 10h 20m
> Remaining Estimate: 0h
>
> Let's suppose we have the following situation:
> - statful ParDo with two timers - timerA and timerB
> - timerA is set for window.maxTimestamp() + 1
> - timerB is set anywhere between <windowStart, windowEnd), let's denote that
> timerB.timestamp
> - input watermark moves to BoundedWindow.TIMESTAMP_MAX_VALUE
> Then the order of timers is as follows (correct):
> - timerB
> - timerA
> But, if timerB sets another timer (say for timerB.timestamp + 1), then the
> order of timers will be:
> - timerB (timerB.timestamp)
> - timerA (BoundedWindow.TIMESTAMP_MAX_VALUE)
> - timerB (timerB.timestamp + 1)
> Which is not ordered by timestamp. The reason for this is that when the input
> watermark update is evaluated, the WatermarkManager,extractFiredTimers() will
> produce both timerA and timerB. That would be correct, but when timerB sets
> another timer, that breaks this.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
