[ 
https://issues.apache.org/jira/browse/BEAM-3776?focusedWorklogId=80099&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-80099
 ]

ASF GitHub Bot logged work on BEAM-3776:
----------------------------------------

                Author: ASF GitHub Bot
            Created on: 13/Mar/18 22:31
            Start Date: 13/Mar/18 22:31
    Worklog Time Spent: 10m 
      Work Description: scwhittle commented on a change in pull request #4793: 
[BEAM-3776] Fix issue with merging late windows where a watermark hold could be 
added behind the input watermark.
URL: https://github.com/apache/beam/pull/4793#discussion_r174305640
 
 

 ##########
 File path: 
runners/core-java/src/test/java/org/apache/beam/runners/core/ReduceFnRunnerTest.java
 ##########
 @@ -873,6 +907,288 @@ public void testWatermarkHoldAndLateData() throws 
Exception {
     tester.assertHasOnlyGlobalAndFinishedSetsFor();
   }
 
+  @Test
+  public void testMergingWatermarkHoldAndLateDataSpecific() throws Exception {
+    LinkedList<LinkedList<Action>> configurations = new LinkedList<>();
+
+    // Simple: late new window
+    LinkedList<Action> actions = new LinkedList<>();
+    actions.add(Action.inputWatermark(40));
+    actions.add(Action.times(1));
+    configurations.add(actions);
+
+    // Simple: late new window, closed and extended.
+    actions = new LinkedList<>();
+    actions.add(Action.inputWatermark(40));
+    actions.add(Action.times(1));
+    actions.add(Action.times(10));
+    configurations.add(actions);
+
+    // Simple: late new window, closed and merged
+    actions = new LinkedList<>();
+    actions.add(Action.inputWatermark(40));
+    actions.add(Action.times(1));
+    actions.add(Action.times(14));
+    actions.add(Action.times(6));
+    configurations.add(actions);
+
+    // Simple: late new window, extended past watermark
+    actions = new LinkedList<>();
+    actions.add(Action.inputWatermark(40));
+    actions.add(Action.times(25));
+    actions.add(Action.times(33));
+    configurations.add(actions);
+
+    // Simple: late new window, extended past watermark, extend more
+    actions = new LinkedList<>();
+    actions.add(Action.inputWatermark(40));
+    actions.add(Action.times(25));
+    actions.add(Action.times(33));
+    actions.add(Action.times(43));
+    configurations.add(actions);
+
+    // Simple: late new window, extended past watermark, extend more
+    actions = new LinkedList<>();
+    actions.add(Action.inputWatermark(40));
+    actions.add(Action.times(25));
+    actions.add(Action.times(33));
+    actions.add(Action.times(11));
+    configurations.add(actions);
+
+    // Simple: new window closes, then extended
+    actions = new LinkedList<>();
+    actions.add(Action.times(11));
+    actions.add(Action.inputWatermark(40));
+    actions.add(Action.times(18));
+    configurations.add(actions);
+
+    // Merging: new window closes, then extended then merged with new window
+    actions = new LinkedList<>();
+    actions.add(Action.times(11));
+    actions.add(Action.inputWatermark(40));
+    actions.add(Action.times(18));
+    actions.add(Action.times(41));
+    actions.add(Action.times(27, 33));
+    configurations.add(actions);
+
+    // Merging: late window, merges with new window
+    actions = new LinkedList<>();
+    actions.add(Action.inputWatermark(40));
+    actions.add(Action.times(29));
+    actions.add(Action.times(41));
+    configurations.add(actions);
+
+    // Merging: late window, new window joined
+    actions = new LinkedList<>();
+    actions.add(Action.inputWatermark(40));
+    actions.add(Action.times(29));
+    actions.add(Action.times(45));
+    actions.add(Action.times(36));
+    configurations.add(actions);
+
+    // Merging: late window, new window all at once
+    actions = new LinkedList<>();
+    actions.add(Action.inputWatermark(40));
+    actions.add(Action.times(29, 45, 36));
+    configurations.add(actions);
+
+    actions = new LinkedList<>();
+    actions.add(Action.inputWatermark(40));
+    actions.add(Action.times(25));
+    actions.add(Action.times(42));
+    actions.add(Action.times(33));
+    configurations.add(actions);
+
+    actions = new LinkedList<>();
+    actions.add(Action.inputWatermark(40));
+    actions.add(Action.times(25));
+    actions.add(Action.times(42));
+    actions.add(Action.times(33, 21));
+    actions.add(Action.inputWatermark(50));
+    actions.add(Action.times(12));
+    configurations.add(actions);
+
+    for (LinkedList<Action> configuration : configurations) {
+      System.out.println("Running config " + configuration.toString());
 
 Review comment:
   Replaced with LOG

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


Issue Time Tracking
-------------------

    Worklog Id:     (was: 80099)
    Time Spent: 1h 50m  (was: 1h 40m)

> StateMerging.mergeWatermarks sets a late watermark hold for late merging 
> windows that depend only on the window
> ---------------------------------------------------------------------------------------------------------------
>
>                 Key: BEAM-3776
>                 URL: https://issues.apache.org/jira/browse/BEAM-3776
>             Project: Beam
>          Issue Type: Bug
>          Components: runner-core
>    Affects Versions: 2.1.0, 2.2.0, 2.3.0
>            Reporter: Sam Whittle
>            Assignee: Sam Whittle
>            Priority: Critical
>          Time Spent: 1h 50m
>  Remaining Estimate: 0h
>
> WatermarkHold.addElementHold and WatermarkHold.addGarbageCollectionHold take 
> to not add holds that would be before the input watermark.
> However WatermarkHold.onMerge calls StateMerging.mergeWatermarks which if the 
> window depends only on window, sets a hold for the end of the window 
> regardless of the input watermark.
> Thus if you have a WindowingStrategy such as:
> WindowingStrategy.of(Sessions.withGapDuration(gapDuration))
>  .withMode(AccumulationMode.DISCARDING_FIRED_PANES)
>  .withTrigger(
>  Repeatedly.forever(
>  AfterWatermark.pastEndOfWindow()
>  .withLateFirings(AfterPane.elementCountAtLeast(10))))
>  .withAllowedLateness(allowedLateness))
> and you merge windows that are late, you might end up holding the watermark 
> until the allowedLateness has passed.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Reply via email to