kafka-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Bhavesh Mistry <mistry.p.bhav...@gmail.com>
Subject Kafka Cluster Monitoring and Documentation of Internals (JMX Metrics) of Rejected Events
Date Mon, 12 Jan 2015 22:04:13 GMT
Hi Kafka Team,

I am trying to find out Kafka Internal and how a message can be corrupted
or lost at brokers side.

I have refer to following documentations for monitoring:
https://cwiki.apache.org/confluence/display/KAFKA/Kafka+Internals
http://kafka.apache.org/documentation.html#monitoring

I am looking at following beans:

"kafka.server":type="BrokerTopicMetrics",name="test-FailedProduceRequestsPerSec"
"kafka.server":type="BrokerTopicMetrics",name="test-BytesRejectedPerSec"


I see following exception on Brokers side rejecting due to large request.
This is great but it does not show the source ip of prodcuer that caused
this issue ? Is there any way to log and capture this ?






*[2014-10-14 22:09:53,262] ERROR [KafkaApi-2] Error processing
ProducerRequest with correlation id 28795280 from client XXXXXon partition
[XXX,17]
(kafka.server.KafkaApis)kafka.common.MessageSizeTooLargeException: Message
size is 2924038 bytes which exceeds the maximum configured message size of
2097152.  *

Can you this be reported as separate metric "MessageSizeTooLargeException
per topic" ?

Also, what is best way to find the CRC check error from the consumer side ?
How do you debug this ?

e.g log line:

*11 Dec 2014 07:22:33,387 ERROR [pool-15-thread-4] *
*kafka.message.InvalidMessageException: Message is corrupt (stored crc =
1834644195, computed crc = 2374999037)*

Also, is there any jira open to update with list all latest metrics and its
format and what it means ?
http://kafka.apache.org/documentation.html#monitoring.  Please see attached
image for list of all metrics.

Version of Broker is 0.8.1.1.

Thanks,

Bhavesh

Mime
  • Unnamed multipart/mixed (inline, None, 0 bytes)
View raw message