The consumers in consumer group 'X' do not have a regex subscription
matching the newly created topic 'C'. They simply subscribe with
the subscribe(java.util.Collection<java.lang.String> topics) method on
topics 'A' and 'B'.

Shouldn't the consumer group have a different state from "Stable" during a
rebalancing regardless of the cause? How else can we determine the consumer
lag of the group during the rebalancing?

Best regards,
Wouter

Have a look at our brand NEW job website: jobs.aca-it.be !


*ACA IT-Solutions NV*
*HQ:* Herkenrodesingel 8B 2.01 | 3500 Hasselt
T +32(0)11 26 50 10 | F +32(0)11 26 50 11
www.aca-it.be | Twitter <https://twitter.com/dorien_jorissen> | Facebook
<http://www.facebook.com/pages/ACA-IT-Solutions/278520212202909> | Linkedin
<http://www.linkedin.com/company/6524>

On 5 February 2018 at 00:13, Hans Jespersen <h...@confluent.io> wrote:

> Do the consumers in consumer group ‘X’ have a regex subscription that
> matches the newly created topic ‘C’?
>
> If they do then they will only discover this new topic once their ‘
> metadata.max.age.ms’  metadata refresh interval has passed, which
> defaults to 5 minutes.
>
> metadata.max.age.ms     The period of time in milliseconds after which we
> force a refresh of metadata even if we haven't seen any partition
> leadership changes to proactively discover any new brokers or partitions
> -hans
>
>
> > On Feb 4, 2018, at 2:16 PM, Wouter Bancken <wouter.banc...@aca-it.be>
> wrote:
> >
> > Hi Hans,
> >
> > Thanks for the response!
> >
> > However, I get this result for all topics, not just for the newly created
> > topic.
> >
> > Situation sketch:
> > 1. I have a consumer group 'X' subscribed to topics 'A' and 'B' with
> > partition assignments and lag information. Consumer group 'X' is
> "Stable".
> > 2a. Topic 'C' is (being) created.
> > 2b. During this creation, I do not have a partition assignment for
> consumer
> > group 'X' for topics 'A' and 'B' but the consumer group is still
> "Stable".
> > 3. A second later: I have a partition assignment for consumer group 'X'
> for
> > topics 'A' and 'B' again and the consumer group is still "Stable".
> >
> > I expected the state of consumer group 'X' during step 2b to be
> > "PreparingRebalance" or "AwaitingSync".
> >
> > Best regards,
> > Wouter
> >
> >> On 4 February 2018 at 21:25, Hans Jespersen <h...@confluent.io> wrote:
> >>
> >> I believe this is expected behavior.
> >>
> >> If there are no subscriptions to a new topic, and therefor no partition
> >> assignments, and definitely no committed offsets, then lag is an
> undefined
> >> concept. When the consumers subscribe to this new topic they may chose
> to
> >> start at the beginning or end of the commit log so the lag cannot be
> >> predicted in advance.
> >>
> >> -hans
> >>
> >>> On Feb 4, 2018, at 11:51 AM, Wouter Bancken <wouter.banc...@aca-it.be>
> >> wrote:
> >>>
> >>> Can anyone clarify if this is a bug in Kafka or the expected behavior?
> >>>
> >>> Best regards,
> >>> Wouter
> >>>
> >>>
> >>> On 30 January 2018 at 21:04, Wouter Bancken <wouter.banc...@aca-it.be>
> >>> wrote:
> >>>
> >>>> Hi,
> >>>>
> >>>> I'm trying to write an external tool to monitor consumer lag on Apache
> >>>> Kafka.
> >>>>
> >>>> For this purpose, I'm using the kafka-consumer-groups tool to fetch
> the
> >>>> consumer offsets.
> >>>>
> >>>> When using this tool, partition assignments seem to be unavailable
> >>>> temporarily during the creation of a new topic even if the consumer
> >> group
> >>>> has no subscription on this new topic. This seems to match the
> >>>> documentation
> >>>> <https://cwiki.apache.org/confluence/display/KAFKA/
> >> Kafka+Client-side+Assignment+Proposal>
> >>>> saying *"Topic metadata changes which have no impact on subscriptions
> >>>> cause resync"*.
> >>>>
> >>>> However, when this occurs I'd expect the state of the consumer to be
> >>>> "PreparingRebalance" or "AwaitingSync" but it is simply "Stable".
> >>>>
> >>>> Is this a bug in the tooling or is there a different way to obtain the
> >>>> correct offsets for a consumer group during a rebalance?
> >>>>
> >>>> I'm using Kafka 10.2.1 but I haven't found any related issues in
> recent
> >>>> changelogs.
> >>>> Best regards,
> >>>> Wouter
> >>>>
> >>
>

Reply via email to