[ 
https://issues.apache.org/jira/browse/CASSANDRA-20202?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17929365#comment-17929365
 ] 

Michael Semb Wever edited comment on CASSANDRA-20202 at 2/22/25 3:22 PM:
-------------------------------------------------------------------------

CI 5.0 "pre-commit" profile:  
[^ci_summary_thelastpickle_mck-samt-20202-5.0_163.html]  
[^results_details_thelastpickle_mck-samt-20202-5.0_163.tar.xz] 

+1


was (Author: michaelsembwever):
CI 5.0 "pre-commit" profile:  
[^ci_summary_thelastpickle_mck-samt-20202-5.0_163.html]  
[^results_details_thelastpickle_mck-samt-20202-5.0_163.tar.xz] 

> Test failure: org.apache.cassandra.transport.SimpleClientPerfTest
> -----------------------------------------------------------------
>
>                 Key: CASSANDRA-20202
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-20202
>             Project: Apache Cassandra
>          Issue Type: Bug
>          Components: Messaging/Client, Test/burn
>            Reporter: Sam Tunnicliffe
>            Assignee: Sam Tunnicliffe
>            Priority: Normal
>             Fix For: 4.1.x, 5.0.x, 5.x
>
>         Attachments: ci_summary_thelastpickle_mck-samt-20202-5.0_163.html, 
> client_capacity_after.png, client_capacity_leak_before.png, 
> lack_of_client_capacity_moment.png, 
> results_details_thelastpickle_mck-samt-20202-5.0_163.tar.xz
>
>
> SimpleClient uses a CQLMessageHandler instance to deal with response messages 
> in the V5 framing format. It looks like CASSANDRA-16663 broke SimpleClient's 
> ability to return capacity allocated during the processing of inbound 
> messages (i.e. responses from server->client) back to the global and 
> per-endpoint resource monitors. The change in behaviour introduced in 
> CASSANDRA-20052, in which we now push back to the client when a capacity for 
> a large message can't be allocated, causes SimpleClientPerfTest in the burn 
> suite to fail on V5/V6. This is purely because the test expects the client to 
> run for a fixed period without interruption, but now it halts as soon as 
> capacity is exhausted. The simple fix is to restore SimpleClient's capacity 
> return.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to