Unable to add a DL380 G5 to SIM 5.1

[NJK-Work]

Hello,

I have a DL380 G5 server (Windows 2003 SP2) running with the latest (7.8) agents and firmware that I am unable to add to SIM.

I should add, that I have been using SIM for years and have added litterally 1000's of devices without problem. So I have gone through all the standard things to look for (SNMP settings, etc.) - I am not trying to brag, but just avoid the basic troubleshooting suggestions that I have already done.

I think the problem is with the SNMP stack itself. When I try to send a test SNMP trap from the "HP Management Agents" tool in the Control Panel, it gives me the following error:

"The trap could not be generated.
Please ensure that the host agent is enabled and running." (screen shot of this error is attached).

I have:
1) Uninstalled the HP agents and reinstalled them.
2) Uninstalled SNMP and reinstalled it.
3) Rebooted countless times.
4) Verified all agents are enabled (except Rack and Cluster - which we don't use) in the "HP Management Agents" tool.
5) Verified all the typical SNMP settings such as security and community name

So I am wondering if anyone has ever seen this and knows how to fix it.

Thanks
Nelson

[loooolaaaaa]

Hi,

Are you trying to add a new server to HP SIM 5.1? is the server pingable from the CMS? If yes, then it will be added to database. Make sue you turn off all the firewall in the managed node and run an Automatic or a Manual discovery.

[Rob Buxton]

That seems like an odd disconnect between the agents and snmp.
Out of interest how was the server built? Smart Start plus a W2K3 SP2 install or an SP1 build plus SP2 upgrade? Or maybe Smart Start wasn't used?
As the above poster suggested - firewalls? but I can't see how that would get in between the agent and snmp. I would have thought that would just stop communication to / from snmp - which would affect discovery but should cause the failure you're seeing.

Any other "interesting" applications on the server that might grab snmp? I know we have a server that has Ericcson software and that seems to grab snmp and won't let it go - makes updating the agents a real pain.

[NJK-Work]

Hi - thanks for the responses.

The server was built from an image of Windows 2003 SP1 (using sysprep). The original image was built from SmartStart (but don't recall the version - I think 7.6). We have other servers built from this image with no problems.

The server was later upgraded to SP2 before being put into production.

The server is not behind a FW.

The server was shipped to a vendor for custom software installation prior to being put into production (this is not standard for us; we usually do everything in house). Past experience tells me that when we have sent servers off to a vendor (again very infrequently) they come back in a questionable state.

The software they installed is pretty unique stuff. I am betting it disrupted SNMP somehow - I just can't figure out how to "reset it". I'll shoot an email off to them to see if they "fess up" to anything.

Thanks
Nelson

[Rob Buxton]

I'm with you on that, I'm always suspicious when a server needs to be sent away for a software product to be installed.

It's obviously not a server you can easily rebuild either!

[Rob Buxton]

Just a thought, try using the Options - Manage System Types and add a new device. You can use this to see if snmp can actually reach the device and then what it returns.

If that cannot get access and you've rechecked snmp you could download and try one of the snmpwalk type tools to see if that can get access.
This should at least narrow down the problem a bit.

[NJK-Work]

Hi Rob,

I tried manually adding the server. It says it is adding it, but never does.

Can you provide any suggestions on an SNMPWalker I should try?

Thanks
Nelson

[David Claypool]

Nelson, with Discovery Filters off, even if SNMP can't be accessed on the device, HP SIM will add a device that responds to ping and then classify it as 'unmanaged.'

[NJK-Work]

Hmmm - I guess I did not try with the filters off. I will try that.

Rob and David - I did find this event in the system log everytime we restart the agents:

Component: Host Agent
Error: Unable to acquire file system information for %1
Cause: This error can be caused by a low memory condition. Rebooting the server may correct this error.

Event ID is 2099 and category is "Host".

(The server has 1 GB of total memory, and from what I can tell seems to be OK in terms of usage).

Does this error/event mean anything to either of you two? Or anyone else?

Thanks
Nelson

[NJK-Work]

Hi,

OK - here is what I did:

1) Tried a free SNMP test tool - from what I can tell, SNMP works fine on the box. I compared the results to a known good box and they box return similar results (I believe it uses system uptime as the check). So I think SNMP is workin.

2) I removed device filters and added the server. It added as David suggested, but is still unmanaged (also as David suggested).

3) I ran an identify on it and got this as part of the results log:

Checking for SNMP support on system.
The system responded to a SNMP request, it supports SNMP....

and later...

Running HP ProLiant management agent identification
System does not have the SNMP ProLiant Server (foundation) agents installed.

But I KNOW the agents are installed and running. Something must be corrupt in them. Since I already uninstalled them and resinstalled them I am not sure what else I can try.

Any suggestions?
Thanks
Nelson