hachikuji commented on code in PR #12372:
URL: https://github.com/apache/kafka/pull/12372#discussion_r911532842
##########
core/src/main/scala/kafka/server/ControllerApis.scala:
##########
@@ -117,6 +117,11 @@ class ControllerApis(val requestChannel: RequestChannel,
s"with context ${request.context}", t)
requestHelper.handleError(request, t)
}
+ } finally {
+ // Only record local completion time if it is unset.
+ if (request.apiLocalCompleteTimeNanos < 0) {
+ request.apiLocalCompleteTimeNanos = time.nanoseconds
Review Comment:
Right. The asynchronous processing time is counted as "remote time." Most of
the time, we are waiting for replication of some kind, so the meaning is mostly
correct. The "local time" is mainly measuring how much time was spent on the
request threads. Perhaps we could consider an additional level of granularity
which accounts for the time spent on the raft/controller threads. Seems like
that would be useful.
--
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: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
