[GitHub] [pulsar] BewareMyPower commented on pull request #13023: Fix consume message order issue when use listener.

[GitBox]

BewareMyPower commented on PR #13023:
URL: https://github.com/apache/pulsar/pull/13023#issuecomment-1090045572

   I've verified the performance regression introduced by this PR. It's a 
really serious bug if we're using `MessageListener`. The affected Pulsar 
versions are:
   - 2.8.3 and later
   - 2.9.1 and later
   - 2.10.0 and later
   
   From my test, the consumer throughput is limited to at most 50+ MiB/s for 
single partition, even if the producer rate was downgraded to only 200 MiB/s. 
Here are my test results for the case that **only 1 partition** is used.
   
   ### Max throughput (10000 MB/s produce rate)
   
   <img width="1251" alt="image" 
src="https://user-images.githubusercontent.com/18204803/161939524-ea2d7867-ef5d-4353-b54c-32a2b8af899c.png";>
   
   <img width="1240" alt="image" 
src="https://user-images.githubusercontent.com/18204803/161939607-f4137e7b-8acd-4bb5-a515-bba12f5d28ac.png";>
   
   ### 200 MB/s produce rate
   
   <img width="1245" alt="image" 
src="https://user-images.githubusercontent.com/18204803/161941047-ca29e34b-9545-4e7f-9cf9-6fff4fd5e481.png";>
   
   <img width="1253" alt="image" 
src="https://user-images.githubusercontent.com/18204803/161941127-82393457-c482-4592-b5a9-c686228610a0.png";>
   
   /cc @merlimat @codelipenghui @dave2wave @Jason918 @eolivelli 


-- 
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.

To unsubscribe, e-mail: commits-unsubscr...@pulsar.apache.org

For queries about this service, please contact Infrastructure at:
us...@infra.apache.org