[jira] [Updated] (BEAM-5063) Watermark does not progress for low traffic streams

["Krzysztof Trubalski (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/BEAM-5063?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Krzysztof Trubalski updated BEAM-5063:
--------------------------------------
    Description: 
We have a Dataflow Job copying data from multiple Kinesis streams into Big Query. Recently we have noticed that the watermark on one of the streams frequently gets stuck although data from that stream is still being processed (it progress only when the traffic increases or Dataflow autoscaling feature kicks in).
  
 Looking at the CloudWatch statistics for the affected stream, it has a really low traffic rate - only ~1 event every few minutes . After investigation and consulting the issue with Google's Dataflow Team, it looks like with such small amount of data on the stream, the function calculating the watermark in KinesisReader reports progress incorrectly.
  
 From my initial investigation, I suspect that the issue might be related to usage of MovingFunction in KinesisReader. In the current implementation, it covers 1 minute period of samples, since obtaining the min value flushes stale values, if the traffic is very low the following call to significance check always returns false (as it relies on the number of samples, and most of them were flushed by get() invocation).
  
  
  

  was:
We have a Dataflow Job copying data from multiple Kinesis streams into Big Query. Recently we have noticed that the watermark on one of the streams frequently gets stuck although data from that stream is still being processed (it progress only when the traffic increases or Dataflow autoscaling feature kicks in).
 
Looking at the CloudWatch statistics for the affected stream, it has a really low traffic rate - only 2 events per minute + 4 events every 5 minutes. After investigation and consulting the issue with Google's Dataflow Team, it looks like with such small amount of data on the stream, the function calculating the watermark in KinesisReader reports progress incorrectly.
 
From my initial investigation, I suspect that the MovingFunction used to keep track of the watermark in KinesisReader is incorrectly used. In the current implementation, min is obtained first and then the significance is checked. Since obtaining the min value flushes stale values, the following call to significance check always returns false (as it relies on the number of samples, and most of them were flushed by get() invocation).
 
 
 


> Watermark does not progress for low traffic streams
> ---------------------------------------------------
>
>                 Key: BEAM-5063
>                 URL: https://issues.apache.org/jira/browse/BEAM-5063
>             Project: Beam
>          Issue Type: Bug
>          Components: io-java-kinesis
>    Affects Versions: 2.5.0
>            Reporter: Krzysztof Trubalski
>            Assignee: Jean-Baptiste Onofré
>            Priority: Major
>
> We have a Dataflow Job copying data from multiple Kinesis streams into Big Query. Recently we have noticed that the watermark on one of the streams frequently gets stuck although data from that stream is still being processed (it progress only when the traffic increases or Dataflow autoscaling feature kicks in).
>   
>  Looking at the CloudWatch statistics for the affected stream, it has a really low traffic rate - only ~1 event every few minutes . After investigation and consulting the issue with Google's Dataflow Team, it looks like with such small amount of data on the stream, the function calculating the watermark in KinesisReader reports progress incorrectly.
>   
>  From my initial investigation, I suspect that the issue might be related to usage of MovingFunction in KinesisReader. In the current implementation, it covers 1 minute period of samples, since obtaining the min value flushes stale values, if the traffic is very low the following call to significance check always returns false (as it relies on the number of samples, and most of them were flushed by get() invocation).
>   
>   
>   



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