Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Threaded View
Kafka, mail # user - Any detail for Kafka Mbeans/yammer-Metrics obtained from JMX


Copy link to this message
-
Re: Any detail for Kafka Mbeans/yammer-Metrics obtained from JMX
Joel Koshy 2013-10-22, 17:58
> (1)"kafka.network":type="RequestChannel",name="Processor-0-ResponseQueueSize",
> "kafka.network":type="RequestChannel",name="Processor-1-ResponseQueueSize",
> "kafka.network":type="RequestChannel",name="Processor-2-ResponseQueueSize"
> Im not getting what  these Processor in (1) are actually for?

Kafka's request handling utilizes a number of processor "threads" and
responses are added to the above queue before sending to the client.
This diagram may help:
https://cwiki.apache.org/confluence/display/KAFKA/Kafka+Internals
> (2)"kafka.server":type="FetcherLagMetrics",name="ReplicaFetcherThread-0-1-host_moni-192-168-145-119-port_9092-demo-0-ConsumerLag"

This reports how far behind that replica is. The format is
ReplicaFetcherThread-<fetcherid>-<leader-broker-id>-<leaderbrokerinfo>-<topic>-<partition>-ConsumerLag

(Fetcher id is typically zero (since the default number of replica
fetchers is one).)

>
> (3)"kafka.server":type="FetcherStats",name="ReplicaFetcherThread-0-1-host_moni-192-168-145-119-port_9092-RequestsPerSec"
> (4)"kafka.server":type="FetcherStats",name="ReplicaFetcherThread-0-2-host_moni-192-168-145-176-port_9092-BytesPerSec"
> The Mbeans (2),(3) and (4) are getting generated on my 3rd broker -
> host_moni-192.168.145.67,kafka running on same port-9092.Any reason for
> these.

Requests per sec/byte rate for replica fetchers, but for all fetches
to the indicated leader brokers (1 and 2).
> I am getting some more Mbeans which are not mentioned here-
> http://kafka.apache.org/documentation.html#monitoring
>
Kafka exposes a *lot* of mbeans - not all of them are listed on that
page that you are referring to. I think the goal of the table was to
provide a minimal set of critical metrics that people may want to set
up monitoring/alerting for.

Thanks,

Joel

>
> On Tue, Oct 22, 2013 at 11:01 PM, Monika Garg <[EMAIL PROTECTED]> wrote:
>
>> Thanks Jun.
>>
>>
>>
>>
>> On Tue, Oct 22, 2013 at 8:51 PM, Jun Rao <[EMAIL PROTECTED]> wrote:
>>
>>> When a broker goes down, we have an optimization in the controller that
>>> proactively shrinks ISR. The ISR shrink rate only measures the rate that
>>> the leader shrinks ISR and doesn't reflect the shrinking by the
>>> controller.
>>> In general, having ISR shrink rate stay at 0 is a good thing since it
>>> means
>>> that the followers can always keep up.
>>>
>>> Thanks,
>>>
>>> Jun
>>>
>>>
>>> On Tue, Oct 22, 2013 at 12:55 AM, Monika Garg <[EMAIL PROTECTED]>
>>> wrote:
>>>
>>> > Thanks a lot Jun.
>>> >
>>> > Now im getting the values in ISR expansion rate,whenever broker goes
>>> down
>>> > and then up.
>>> >
>>> > But still ISR shrink rate is always having value 0.
>>> >
>>> > what I have done is:
>>> >
>>> > I made 3 node cluster.Than created 5 partitions with replica=3.
>>> > partitions got distributed on 1,2,3 brokers with isr= 1,2,3, and
>>> replica =
>>> > 1,2,3
>>> > broker -3 is leader for 2 partitions.
>>> >
>>> > then i stopped one broker(say 3),then isr = 1,2 and replica = 1,2,3.
>>> > In this case I waited for some time to get the ISR shrink value to get
>>> > updated in my jconsole,but value remained 0 for all attributes of ISR
>>> > shrink rate.
>>> >
>>> > Then I started broker(say 3) that i stopped earlier,what I saw is:
>>> >
>>> > ISR expansion rate has the values as:
>>> >
>>> > count -2 (on both other brokers 1,2)
>>> > oneminuterate - 0.012 etc.
>>> >
>>> > then I again stopped broker-1,still isr shrink rate not getting any
>>> > value.Can you please check?
>>> >
>>> > One more thing if you can please have a look:
>>> > How the other attributes like FifteenMinuteRate,OneMinuteRate etc are
>>> > continuosly changing for any Mbean.
>>> >
>>> >
>>> >
>>> >
>>> >
>>> >
>>> >
>>> > On Tue, Oct 22, 2013 at 9:00 AM, Jun Rao <[EMAIL PROTECTED]> wrote:
>>> >
>>> > > Updated the description in
>>> > > http://kafka.apache.org/documentation.html#monitoring. Does that help
>>> > make
>>> > > things clearer?
>>> > >
>>> > > Thanks,
>>> > >
>>> > > Jun
>>> > >
>>> > >
>