Along these lines is there a list of JMX beans and what values are exposed
w/some description of the metric they communicate?
I am trying to understand how the system is doing and don't know what I don't
know yet :)
On Aug 2, 2012, at 11:20 AM, Joseph Boyd wrote:
> Thanks! I was (wrongly) looking for this information on the kafka
> process, and not on the consumer process. Once I looked there I found
> kafka.ConsumerStats.
>
> (still not sure how I read dates wrong for the JIRA, but that happens)
>
>
> ...joe
>
>
> On Wed, Aug 1, 2012 at 10:22 PM, Jun Rao <jun...@gmail.com> wrote:
>> Joe,
>>
>> It should be there. 0.7.1 was released last month and that jira was closed
>> last year.
>>
>> Thanks,
>>
>> Jun
>>
>> On Wed, Aug 1, 2012 at 10:51 AM, Joseph Boyd <joseph.b...@cbsinteractive.com
>>> wrote:
>>
>>> It looks like kafka.ConsumerStats isn't in 0.7.1, which is why I'm not
>>> seeing it.
>>>
>>> My read of JIRA / mailing list archives shows KAFKA-136 went in
>>> shortly after the 0.7.1 release was rolled up.
>>>
>>>
>>> Thanks for the pointer, though.
>>>
>>>
>>> ...joe
>>>
>>>
>>>
>>>
>>> On Wed, Aug 1, 2012 at 10:19 AM, Jun Rao <jun...@gmail.com> wrote:
>>>> Joe,
>>>>
>>>> We have the following bean name and type registered for consumers. There
>>> is
>>>> also a CheckOffsetLag tool.
>>>>
>>>> kafka:type=kafka.ConsumerStats
>>>> /**
>>>> * JMX interface for monitoring consumer
>>>> */
>>>> trait ZookeeperConsumerConnectorMBean {
>>>> def getPartOwnerStats: String
>>>> def getConsumerGroup: String
>>>> def getOffsetLag(topic: String, brokerId: Int, partitionId: Int): Long
>>>> def getConsumedOffset(topic: String, brokerId: Int, partitionId: Int):
>>>> Long
>>>> def getLatestOffset(topic: String, brokerId: Int, partitionId: Int):
>>> Long
>>>> }
>>>>
>>>> Thanks,
>>>>
>>>> Jun
>>>>
>>>> On Wed, Aug 1, 2012 at 10:10 AM, Joseph Boyd <
>>> joseph.b...@cbsinteractive.com
>>>>> wrote:
>>>>
>>>>> Hi there,
>>>>>
>>>>> Is there any way to know from kafka's JMX interface whether there are
>>>>> unread messages on a topic? I think the answer is 'no', based on my
>>>>> read of documentation (esp. the 'high water mark'), and my own
>>>>> experiments with producing test messages, looking at the stats, and
>>>>> then consuming messages (both with/without committing messages
>>>>> consumed).
>>>>>
>>>>> I thought I'd double check on the users list anyhow, though, in case
>>>>> I'm missing something. The KAFKA-260 work looks useful in this area
>>>>> too.
>>>>>
>>>>> Thanks for any help.
>>>>>
>>>>>
>>>>> ...joe
>>>>>
>>>
