strange rrd results via ncv

["Whilding, Craig" <Craig_Whilding_TERM (at) relay1.mentorg.com>]

We recently rolled out a sun grid engine platform and I'm now trying to
gather performance/usage information from it (if someone already has a
collection of scripts this would also help me immensely).

 

I've written a perl script that passes job queue information to hobbit
which outputs in the following format:

 

Running    jobs : 7
Queued     jobs : 0
Xfering    jobs : 0
Errorstate jobs : 0

 

This should be ideal for ncv and I have therefore added ginfo=ncv to
TEST2RRD in hobbitserver.cfg as well as
NCV_qinfo="Runningjobs:GAUGE,Queuedjobs:GAUGE,Xferingjobs:GAUGE,Errorsta
tejobs:GAUGE" on its own line because this is fluctuating data and not a
counter.

 

The graph definition looks like this:

[qinfo]

        TITLE Q Info

        YAXIS Jobs

        DEF:running=qinfo.rrd:Runningjobs:AVERAGE

        DEF:queued=qinfo.rrd:Queuedjobs:AVERAGE

        DEF:xfering=qinfo.rrd:Xferingjobs:AVERAGE

        DEF:errorstate=qinfo.rrd:Errorstatejobs:AVERAGE

        LINE2:running#0000FF:Running Jobs

        LINE2:queued#00CC33:Queued Jobs

        LINE2:xfering#FF0000:Xfering Jobs

        LINE2:errorstate#FFFF00:Errorstate Jobs

        COMMENT:\n

        GPRINT:running:LAST:Running jobs \: %5.1lf%s (cur)

        GPRINT:running:MAX: \: %5.1lf%s (max)

        GPRINT:running:MIN: \: %5.1lf%s (min)

        GPRINT:running:AVERAGE: \: %5.1lf%s (avg)\n

        GPRINT:queued:LAST:Queued jobs \: %5.1lf%s (cur)

        GPRINT:queued:MAX: \: %5.1lf%s (max)

        GPRINT:queued:MIN: \: %5.1lf%s (min)

        GPRINT:queued:AVERAGE: \: %5.1lf%s (avg)\n

        GPRINT:xfering:LAST:Xfering jobs \: %5.1lf%s (cur)

        GPRINT:xfering:MAX: \: %5.1lf%s (max)

        GPRINT:xfering:MIN: \: %5.1lf%s (min)

        GPRINT:xfering:AVERAGE: \: %5.1lf%s (avg)\n

        GPRINT:errorstate:LAST:Errorstate jobs \: %5.1lf%s (cur)

        GPRINT:errorstate:MAX: \: %5.1lf%s (max)

        GPRINT:errorstate:MIN: \: %5.1lf%s (min)

        GPRINT:errorstate:AVERAGE: \: %5.1lf%s (avg)\n

 

 

As you can see from the attached graph there are some rather strange
results coming out and the rrd file contains values such as 3.9/4.8 when
you really can't have .x of a job and the output from the script only
ever shows whole numbers as it should.

 

Thanks for any help with what's going wrong.

 

 

Regards,

Craig Whilding

IT Systems Contractor

Mentor Graphics