Home

Welcome to the jmxxymon wiki!

This is a rather simple program for extracting JMX information and turning it into xymon/hobbit/big brother data.
From there you can then do things like:
- Check a service is up
- Check that a service is receiving/sending messages
- Draw graphs of internal information
- etc.

Usage

To use with the standard jmx java.lang exports, follow the steps below. This will give you graphs of cpu time and memory usage. Obviously any other figure exported by jmx can be got at in a similar way. The lessThan test will trigger a red light if memory use exceeds 10M. Probably way too low, but handy if you want to prove that it works.

Create config files in xymon/client/etc/jmx
- vm.conf

vm,cpuTime,@port@,java.lang,OperatingSystem,ProcessCpuTime
vmmem,memInit,@port@,java.lang,Memory,HeapMemoryUsage.init
vmmem,memCommitted,@port@,java.lang,Memory,HeapMemoryUsage.committed
vmmem,memMax,@port@,java.lang,Memory,HeapMemoryUsage.max
vmmem,memUsed,@port@,java.lang,Memory,HeapMemoryUsage.used,lessThan(10000000)

- Add some lines in bb-hosts as script fodder

0.0.0.0 tomcat # noconn jmxhost=localhost jmxport=1070 \
	       	 	jmx=jmx/vm.conf   	       

- Add a hobbitlaunch entry for the monitor

[jmxmonitor]
	ENVFILE /home/pab/xymon/client/etc/hobbitclient.cfg
	CMD $BBHOME/scripts/jmx_wrapper.sh
	LOGFILE $BBSERVERLOGS/jmx.log
	INTERVAL 30s

jmx_wrapper.sh is in the doc directory, for no good reason.

- Add some graph type stuff to hobbitserver.cfg (or in /etc/xymon/hobbitserver.d). Here I go for cpuTime and memory use. You might want thread counts, or whatever.

TEST2RRD="$TEST2RRD,vm=ncv"
SPLITNCV_vm="cpuTime:DERIVE,loadAvg:GAUGE"

TEST2RRD="$TEST2RRD,vmmem=ncv"
SPLITNCV_vmmem="memUsed:GAUGE,memCommitted:GAUGE,memMax:GAUGE"

- and a graph definition to hobbitgraph.cfg

[vm]
	TITLE Java Cpu Time
	YAXIS Time
	DEF:cpuTime=vm,cpuTime.rrd:lambda:AVERAGE
	LINE:cpuTime#000000:Cpu Time
	GPRINT:cpuTime:LAST: \: %5.1lf (cur)

[vmmem]
	TITLE Java Virtual Memory
	YAXIS Time
	DEF:memMax=vmmem,memMax.rrd:lambda:AVERAGE
	DEF:memUsed=vmmem,memUsed.rrd:lambda:AVERAGE
	DEF:memComm=vmmem,memCommitted.rrd:lambda:AVERAGE
	CDEF:memPercent=memUsed,100,*,memMax,/
	CDEF:memCPercent=memComm,100,*,memMax,/
	LINE:memPercent#000000:Mem Used
	LINE:memCPercent#000000:Mem Committed\n
	GPRINT:memUsed:LAST:Used \: %5.1lf
	GPRINT:memPercent:LAST:(%0.2lf%%) \n
	GPRINT:memComm:LAST:Committed \: %5.1lf
	GPRINT:memCPercent:LAST:(%0.2lf%%)	\n
	GPRINT:memMax:LAST:Max \: %5.1lf	\n

And you get graphs of your tomcat mem usage and cpu.