Thanks Jun and Chia for review and feedback. Initially I thought this might be a breaking change for someone who has written a transformation over topic name and expects them to be "_" replaced, if dot exists. But the fix is actually a correction of the topic name itself hence was wondering if we should just correct it without deprecating existing. Unlike KIP-773 where we required 'ns' per other metrics naming convention for data captured, here we are correcting a tag value.
In case we want to deprecate then there could be 2 options I was thinking of: 1. As the metric names seem to be appropriate hence add another tag/label as "fetch-topic" which should contain the topic name without any replacement. This approach will not change the existing "topic" tag data and in the subsequent releases we can drop this "topic" tag as well. 2. Deprecate old metrics and introduce new ones - I find the existing consumer metrics <https://kafka.apache.org/20/generated/consumer_metrics.html> named appropriately, hence finding it hard to come up with better names for new metrics, any suggestions if we go with this approach? Regards, Apoorv Mittal On Wed, Nov 20, 2024 at 5:57 PM Jun Rao <j...@confluent.io.invalid> wrote: > Hi, Chia-Ping, > > We could deprecate those metrics too. Then, we probably want to do a formal > KIP on this. > > Thanks, > > Jun > > On Tue, Nov 19, 2024 at 3:11 AM Chia-Ping Tsai <chia7...@gmail.com> wrote: > > > hi Apoorv, > > > > Should we follow the deprecation cycle in this KIP? I mean, we should > keep > > both metrics for now and add a description to inform users that the > metrics > > are deprecated. > > > > hi Jun > > > > The incorrect topic metrics have existed for a while, so I believe it > would > > be better to follow a deprecation cycle, similar to KIP-773. ( > > > > > https://cwiki.apache.org/confluence/display/KAFKA/KIP-773%3A+Differentiate+consistently+metric+latency+measured+in+millis+and+nanos > > ) > > > > Best, > > Chia-Ping > > > > Jun Rao <j...@confluent.io.invalid> 於 2024年11月19日 週二 上午3:48寫道： > > > > > Hi, Apoorv, > > > > > > We did the dot replacement in Yammer metric scope to help with the > naming > > > schema used in Graphite and the java consumer code just copied it. > Since > > > this is an issue very specific to Graphite, it's better not to change > the > > > name directly in Kafka metric. If needed, a Graphite specific reporter > > > could be written to address its specific issue. I agree that this is > more > > > like a bug than a new feature. We could probably just fix the issue in > > the > > > java consumer without a KIP. > > > > > > Thanks, > > > > > > Jun > > > > > > On Fri, Nov 15, 2024 at 2:37 PM Apoorv Mittal < > apoorvmitta...@gmail.com> > > > wrote: > > > > > > > Moreover on a second thought, as this should be a correction in > emitted > > > > topic names then we should not require a KIP per say but just fix it > as > > > > part of jira itself. Please let me know your thoughts. > > > > > > > > Regards, > > > > Apoorv Mittal > > > > > > > > > > > > > > > > On Fri, Nov 15, 2024 at 10:16 PM Apoorv Mittal < > > apoorvmitta...@gmail.com > > > > > > > > wrote: > > > > > > > > > Hi All, > > > > > I have added the Context section in the KIP as per the discussion > on > > > the > > > > > jira <https://issues.apache.org/jira/browse/KAFKA-12469>. Please > let > > > me > > > > > know if there is any feedback or I shall directly start the voting > > > > thread. > > > > > > > > > > Regards, > > > > > Apoorv Mittal > > > > > > > > > > > > > > > On Tue, Nov 12, 2024 at 12:49 PM Apoorv Mittal < > > > apoorvmitta...@gmail.com > > > > > > > > > > wrote: > > > > > > > > > >> Hi All, > > > > >> I would like to start a discussion on KIP-1109: > > > > >> > > > > > > > > > > https://cwiki.apache.org/confluence/display/KAFKA/KIP-1109%3A+Unifying+Kafka+Consumer+Topic+Metrics > > > > >> > > > > >> This KIP streamlines topic and topic-partition metrics for Kafka > > > > >> Consumer, emitting the user defined topic name (as like > > > kafka-producer). > > > > >> > > > > >> Regards, > > > > >> Apoorv Mittal > > > > >> > > > > > > > > > > > > > > >
