Funny you mention that, I just completed a script that does disk iostat, and
graphs all the columns from an iostat -X
(Solaris centric, but I am sure you could mod it for any other OS)
The graphs show up in trends.
For us it made no sense to create an alert based on iostat, but rather to
record it for and use it as a forensic tool if/when things went bad.
It has been running in test overnight, and if all looks good, I will post it
later today.
Cheers
Vernon
On Thu, Sep 9, 2010 at 4:47 AM, Daniel Bourque <user-a141068964db@xymon.invalid>wrote:
Hi all,
I want to create graphs for i/o per seconds on all disks on host
bob.somewhere.com . I am trying to do like the disk check, but I can't
hobbit to create multiple rrd files.
so my check ends up running the following
/usr/lib/hobbit/client/bin/bb 1.1.1.1 "status
bob,somewhere,com.iostat_tps,sda green Wed Sep 8 15:01:49 CDT 2010
iostat_tps : 23.40
"
/usr/lib/hobbit/client/bin/bb 1.1.1.1 "status
bob,somewhere,com.iostat_tps,sdb green Wed Sep 8 15:01:49 CDT 2010
iostat_tps : 12.00
"
in hobbitserver.cfg :
I appended "iostat_tps=ncv" to TEST2RRD
I added NVC_iostat_tps="iostat_tps:GAUGE"
----
I restarted the rrd processes
----
I am not seeing
iostat_tps,sda.rrd
iostat_tps,sdb.rrd
under /opt/hobbit/data/rrd/bob.somewhere.com/
----
my hope is to able to do a graph using something like
[iostat_tps]
FNPATTERN iostat_tps(.*).rrd
Any ideas ?
hobbit 4.2.0
RHEL 3
Thank !
--
Dan