Agentless? That's funny. That's OK for Windows, which seems to like
sending everything and the kitchen sink out in SNMP, but have you
tried monitoring Solaris with SNMP? Consider Sparc hardware,
running LDOMs, with zones in the LDOMs. (Not an unusual mix - I
have a number of clients doing this) I have yet to see an
"auto-discovery" monitoring tool come even close to getting it
right. Disks, most times, but for anything else beyond the
"standard" stuff, forget it. I did witness one priceless example
where one of these "auto-discovery" monitors announced, and started
monitoring and graphing an MS-SQL database it discovered on a
Solaris Sparc server. Sales guy left with his tail between his
legs.
I set up Xymon a few years back at a local oil and gas company. A
few hundred nodes, a healthy mix of Solaris, Linux and Windows,
filers, mass storage, network devices, apps monitoring, database
monitoring, the works. Besides the usual Xymon interfaces, I had
also set up specific pages to provide application-centric views,
allowing application custodians to monitor a single page giving
them a complete view of the health of their application with
associated dependencies listed on their page. Since I left, nearly
three years ago now, they have tried (more for political reasons
than technical) to replace Xymon three times. They still use Xymon
:-)
I have yet to see agentless work properly.
Just my tuppence worth.
Regards Vernon
On 13 June 2012 11:03, Ralph Mitchell <user-00a5e44c48c0@xymon.invalid <mailto:user-00a5e44c48c0@xymon.invalid>> wrote:
I took a quick look at the Zenoss community page, and fairly
quickly noticed one interesting thing - apparently it's agentless.
No agent to install on target systems, because it picks up
information using SNMP. That's not necessarily bad, but some
companies don't like it. Or you can use "zencommand" to run nagios
binaries on your target. That info is dated May 2008, so maybe it
grew up since?
Right now I'm running a somewhat modified xymon client out of cron on AIX, RHEL and SuSE systems, delivering reports over https on
port 443 (which is open anyway) instead of opening port 1984. It
doesn't use any compiled binaries other than the standard system
commands, which means that I don't have to recompile for system
updates. Bourne shell all the way... :-)
My situation may be unique, though - we have to comply with both
PCI *and* DIACAP requirements.
Ralph Mitchell
On Tue, Jun 12, 2012 at 8:45 PM, Jones, Elizabeth <user-f0dffc744f19@xymon.invalid <mailto:user-f0dffc744f19@xymon.invalid>> wrote:
Would any of you mind telling me why you use xymon? I have a co-worker who is pushing hard for zenoss, and his arguments include
?gui discovery tool?, ?no hand editing of config files?, ?it is an
enterprise level monitoring system and you can buy support and
bring in a consultant to set it up for you?, and the ever popular
?everyone is using zenoss now?. ____
__ __
Elizabeth Jones____
__ __
-- "While it is futile to try to eliminate risk, and questionable
to try to minimize it, it is essential that the risks taken be the
right risks. " - Peter F. Drucker
This body part will be downloaded on demand.