[GitHub] [kafka] divijvaidya commented on pull request #13798: KAFKA-15028: AddPartitionsToTxnManager metrics

2023-06-28 Thread via GitHub
divijvaidya commented on PR #13798: URL: https://github.com/apache/kafka/pull/13798#issuecomment-1611605733 > so only that one verification per transaction ah, I had missed this part that it will not be recorded on "every" message append. Only for verified cases. I think we should be

[GitHub] [kafka] divijvaidya commented on pull request #13798: KAFKA-15028: AddPartitionsToTxnManager metrics

2023-06-27 Thread via GitHub
divijvaidya commented on PR #13798: URL: https://github.com/apache/kafka/pull/13798#issuecomment-1609922486 > if this does affect performance too much Maybe we could use some type of recoding level for Yammer metrics too? We already have a configuration at:

[GitHub] [kafka] divijvaidya commented on pull request #13798: KAFKA-15028: AddPartitionsToTxnManager metrics

2023-06-27 Thread via GitHub
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