I do know that Sun supply some KM's to monitor their own hardware, which we have used in the past. They are rather pricey though!
The way we are looking to monitor our Sun hardware is via SNMP. We have loaded the MIB into our PEM Connect/SNMP software and are having the alerts sent this way. I'm unsure as to how easy it would be to write your own KM to simply intercept the SNMP traps, but I'm sure someone out there has done it.
HSBC Bank Plc
Thank you Jason. You mean use Patrol Enterprise Manager to do the hardware via SNMP and Patrol KM to do other stuff? And Do you use POM in your env.? Guess not. WE use Patrol to POM then POM to PEM.
I wrote a similar KM for a customer. Since you can obtain this SunE10K info command line it is possible to incorporate it into a KM. My customer only wanted little pieces of info, which is why they opted for the custom KM versus the Full Blown one from Sun
I believe the Sun E10K KM is well expensive.
However it works quite simply. I think it walks the MIB tree rather than receive SNMP traps.
I traversed the MIB tree from a remote box using HPs NNM snmp command line utilities. The same functionality lies within PSL with the snmp()functions.
I think by walking the MIB sections you're interested concerned with in your PSL code you can populate the parameters to reflect the hardwares status. The collector doesn't have to be on the actual node in question.
SUNs own E10K KM does not sit on the SSP nodes
I just finished writing a KM that monitors HP-UX disks, and it was quite easy. For system disks, I am running ioscan -FC disk and searching for anything that is not "CLAIMED" in the status field. I use the device path as the instance name.
For AutoRaid, Nike10, and Nike20's, I pipe a script through cstm and grep for the word "FAULT DETECTED."
I read some speak of SNMP, but have not researched what I can pull from the daemon running on an HP machine. Does anyone know? Can I get easier information for disks, nic cards, tape drives, etc from this??
I have the ESP km (v 2.6) , and I must say it is not worth the purchase , only if you are using 4500's or up do you get any usefull info such as PSU status etc .. , it is very dependant on the server model as to what info you will get from it .
I have it one some servers where all I get is CPU and Motherboard info , and not much of it at that , there is also a fair bit on config required for this KM , you have to install the ESD agent , which is part of the Sun Mgt Centre , blah blah blah.....
Overall just aint woth the effort in my view.
and dont even get me started on the SC3 km !!!
We have been using the E10K CRI* KM's on the management stations. And they indeed perform an SNMPwalk across certain OID.
But because this KM was too CPU intensive it has been disabled.
Now we're investigating other solution for monitoring our new Sun F15K's.
And found several possibilities:
- Install a similar KM on every 15K, and collect the info via SNMP or via the SunMC agent (not available, so needs development).
- Use the Sun Management Center - Patrol integration,where there is a Patrol Agent installed on the Sun MC, and the SunMC will forward the alerts it receives from all the SunMC agents to the Patrol Agent. From there it can be picked up by PEM. (available now, but low level of integration = Patrol console part is weak)
- Use the PEM/Connect for SunMc, to get the alert from SunMC directly into PEM, without going through the Patrol Agent. (not out yet, should be available by the end of the year)
Do you remmember which release of the E10K km version that you were using? Been looking into getting some hardware info from our E10K's but don't want to use something that may cause problems.
We are also running Sun KMs for Netra's and E3500. We tried getting the SunFIRE KM working but had so many issues that we disabled it. Very pricey for not much gain on some of the servers. We disabled the use of this KM on our e420r's
Eventually we jut wrote our own KM, with only the parameters we wanted and saved some money ...
i have to agree with Xpericon that that is the best solution , do something yourself , we have now cancelled our maintenance on the ESP km and removed it as we found it was giving us more grief than benefit, we had major problems with the stability of the ESD agent (Component of SunMC)
could have picked half the stuff up it was giving us just from var/adm/messages anyway.
P.S Has anyone else noticed that SUN are not in the partner forum section anymore ? , anyone know why ?