divijvaidya commented on PR #13798: URL: https://github.com/apache/kafka/pull/13798#issuecomment-1609327501
Another point I want to call out is that Yammer metrics histogram is notorious for consuming CPU (and increase latency). It consumes ~4-5% CPU on the network threads for calculating per request latency. Given that we are doing this on the latency critical append path (right?), I suspect we may encounter some latency increase + CPU increase. Although one histogram calculation here should be ok, but it would be nice if you get some producer-perf.sh data in as well to ensure that this metric isn't adversely impacting latency. Thoughts? -- 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: jira-unsubscr...@kafka.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
