[jira] [Commented] (RATIS-1513) Streaming tests may fail with TimeoutIOException: Timeout 3000ms

["GuoHao (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/RATIS-1513?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17487944#comment-17487944 ] 

GuoHao commented on RATIS-1513:
-------------------------------

[~szetszwo] , 

I suspect that it is not the timeout handling bug, but the logic that takes too long to process. Because it's an incidental problem.

 

In the processing logic, I printed the log at various points, with `DataStreamManagement#startTransaction` taking the longest.

 

!image-2022-02-07-16-10-37-396.png!

 

It sends raft Async RPC to get raft log updates so statemachine.link can get logEntry.

And ozone uses the index in logEntry as BCSID. Is it possible to increase efficiency by having link enabled as an option and ozone otherwise fetching a unique ID as the BCSID?

> Streaming tests may fail with TimeoutIOException: Timeout 3000ms
> ----------------------------------------------------------------
>
>                 Key: RATIS-1513
>                 URL: https://issues.apache.org/jira/browse/RATIS-1513
>             Project: Ratis
>          Issue Type: Sub-task
>          Components: Streaming, test
>            Reporter: Tsz-wo Sze
>            Assignee: Tsz-wo Sze
>            Priority: Major
>         Attachments: image-2022-02-07-16-10-37-396.png
>
>
> Recently, the Streaming unit tests might fail with TimeoutIOException: Timeout 3000ms.  The failure seems not specific to any tests.  Below is an example.
> {code}
> Error:  testMultipleStreamsMultipleServersStepDownLeader(org.apache.ratis.datastream.TestNettyDataStreamStarTopologyWithGrpcCluster)  Time elapsed: 34.006 s  <<< ERROR!
> 544
> java.util.concurrent.CompletionException: org.apache.ratis.protocol.exceptions.TimeoutIOException: Timeout 3000ms: Failed to send DataStreamWindowRequest:seqNum=12,DataStreamRequestHeader:clientId=client-0E50888F0908,type=STREAM_DATA,id=404,offset=7441432,length=0
> 545
> {code}



--
This message was sent by Atlassian Jira
(v8.20.1#820001)