In our case, jmxtrans returns: ReplicaManager.ISRShrinksPerSec.Count ReplicaManager.ISRShrinksPerSec.MeanRate
ReplicaManager.ISRExpandsPerSec.OneMinuteRate ReplicaManager.ISRExpandsPerSec.Count ReplicaManager.ISRExpandsPerSec.MeanRate ReplicaManager.ISRExpandsPerSec.OneMinuteRate ReplicaManager.LeaderCount.Value ReplicaManager.PartitionCount.Value In other words, the bean name is include in the metric name when jmxtrans sends it out. In our case, we are using graphite. Not sure if that makes a difference. Paul On 11/5/13 12:59 AM, "Priya Matpadi" <priya.matp...@ecofactor.com> wrote: >Paul, >Thank you! I didn't realize the quotes are important to keep. >I have question on name=\"*\"" > >For the type=ReplicaManager, there are at least 4 Mbean names >"kafka.server":name="PartitionCount",type="ReplicaManager" >"kafka.server":name="LeaderCount",type="ReplicaManager" > >"kafka.server":name="ISRShrinksPerSec",type="ReplicaManager" >"kafka.server":name="ISRExpandsPerSec",type="ReplicaManager" > >In ganglia metrics dropdown, I only see 4 elements as configured above >i.e. > >ReplicaManager.Count >ReplicaManager.OneMinuteRate >ReplicaManager.MeanRate >ReplicaManager.Value > >Then, how can one tell if for e.g. Value pertains to LeaderCount or >PartitionCount? >Even though it is laborious, isn't it good idea to explicitly specify the >name? > >Thanks, >Priya > > > > >On Mon, Nov 4, 2013 at 5:10 PM, Paul Mackles <pmack...@adobe.com> wrote: > >> It looks like you are missing quotes in the object name. Here is a >>snippet >> from our jmxtrans configs: >> >> "resultAlias": "ReplicaManager", >> "obj": "\"kafka.server\":type=\"ReplicaManager\",name=\"*\"", >> "attr": [ >> "Count", >> "OneMinuteRate", >> "MeanRate", >> "Value" >> ] >> >> >> Unless more recent versions of kafka get rid of the quotes (we are on an >> older 0.8 version still). >> >> Paul >> >> On 11/4/13 8:02 PM, "Priya Matpadi" <priya.matp...@ecofactor.com> wrote: >> >> >I am trying to send kafka metrics for display to ganglia server using >> >latest download from https://github.com/adambarthelson/kafka-ganglia. >> > >> >Here's my kafka_metrics.json >> >{ >> > "servers" : [ { >> > "port" : "9999", >> > "host" : "ecokaf1", >> > "queries" : [ { >> > "outputWriters" : [ { >> > "@class" : >> >"com.googlecode.jmxtrans.model.output.GangliaWriter", >> > "settings" : { >> > "groupName" : "kafka stats 2", >> > "port" : 8649, >> > "host" : "ecokaf1" >> > } >> > } ], >> > "obj" : "kafka.server:type=ReplicaManager,name=PartitionCount", >> > "resultAlias": "Kafka.ReplicaManager", >> > "attr" : [ "Value" ] >> > } ], >> > "numQueryThreads" : 2 >> > } ] >> >} >> > >> >I start jmxtrans as follows: >> >sudo ./jmxtrans.sh start kafka_metrics.json >> > >> > >> >JMXTrans starts without any issues. Here are the logs: >> > >> >[05 Nov 2013 00:49:48] [main] 0 INFO >> >(com.googlecode.jmxtrans.JmxTransformer:134) - Starting Jmxtrans on : >> >kafka_metrics.json >> >[05 Nov 2013 00:49:48] [main] 386 DEBUG >> >(com.googlecode.jmxtrans.JmxTransformer:354) - Loaded file: >> >/usr/share/jmxtrans/kafka_metrics.json >> >[05 Nov 2013 00:49:48] [main] 392 DEBUG >> >(com.googlecode.jmxtrans.model.output.GangliaWriter:119) - Validated >> >Ganglia metric [host: ecokaf1, port: 8649, addressingMode: UNICAST, >>ttl: >> >5, >> >v3.1: true, units: '', slope: BOTH, tmax: 60, dmax: 0, spoofedHostName: >> >192.168.3.1:ecokaf1, groupName: 'kafka stats 2'] >> >[05 Nov 2013 00:49:48] [main] 406 DEBUG >> >(com.googlecode.jmxtrans.JmxTransformer:429) - Scheduled job: >> >ecokaf1:9999-1383612588479-3580845919 for server: Server [host=ecokaf1, >> >port=9999, url=null, cronExpression=null, numQueryThreads=2] >> >[05 Nov 2013 00:49:48] [ServerScheduler_Worker-1] 414 DEBUG >> >(com.googlecode.jmxtrans.jobs.ServerJob:31) - +++++ Started server job: >> >Server [host=ecokaf1, port=9999, url=null, cronExpression=null, >> >numQueryThreads=2] >> >[05 Nov 2013 00:49:48] [ServerScheduler_Worker-1] 544 DEBUG >> >(com.googlecode.jmxtrans.util.JmxUtils:102) - ----- Creating 1 query >> >threads >> >[05 Nov 2013 00:49:48] [ServerScheduler_Worker-1] 553 DEBUG >> >(com.googlecode.jmxtrans.jobs.ServerJob:50) - +++++ Finished server >>job: >> >Server [host=ecokaf1, port=9999, >> >url=service:jmx:rmi:///jndi/rmi://ecokaf1:9999/jmxrmi, >> >cronExpression=null, >> >numQueryThreads=2] >> >[05 Nov 2013 00:50:48] [ServerScheduler_Worker-2] 60401 DEBUG >> >(com.googlecode.jmxtrans.jobs.ServerJob:31) - +++++ Started server job: >> >Server [host=ecokaf1, port=9999, >> >url=service:jmx:rmi:///jndi/rmi://ecokaf1:9999/jmxrmi, >> >cronExpression=null, >> >numQueryThreads=2] >> >[05 Nov 2013 00:50:48] [ServerScheduler_Worker-2] 60404 DEBUG >> >(com.googlecode.jmxtrans.util.JmxUtils:102) - ----- Creating 1 query >> >threads >> >[05 Nov 2013 00:50:48] [ServerScheduler_Worker-2] 60410 DEBUG >> >(com.googlecode.jmxtrans.jobs.ServerJob:50) - +++++ Finished server >>job: >> >Server [host=ecokaf1, port=9999, >> >url=service:jmx:rmi:///jndi/rmi://ecokaf1:9999/jmxrmi, >> >cronExpression=null, >> >numQueryThreads=2] >> > >> >Does any one spot an issue with the json? >> > >> >When I add the jvmheapmemory and jvmGC examples to my json, jmxtrans >>sends >> >those metrics to Ganglia. Corresponding rrds files are created in >>ganglia >> >location and I can see the following in jmxtrans.log: >> >[05 Nov 2013 00:45:44] [pool-30-thread-1] 1740436 DEBUG >> >(com.googlecode.jmxtrans.model.output.GangliaWriter:141) - Sending >>Ganglia >> >metric heap.HeapMemoryUsage_init=263441792 >> >[05 Nov 2013 00:45:44] [pool-30-thread-2] 1740449 DEBUG >> >(com.googlecode.jmxtrans.model.output.GangliaWriter:141) - Sending >>Ganglia >> >metric GC.PSScavenge.CollectionCount=334 >> > >> >And I know that kafka is publishing jmx metrics because I can see them >> >using jmxsh-R5 command line tool, >> > >> >Appreciate any pointers you can provide with regards to kafka. >> >Thanks, >> >Priya >> >>
