DL380's with ILO - SNMP traps cpqSm2ServerPowerOutage & cpqSm2ServerReset detected

[matt b_1]

Hi,

We have DL380 G6 servers running load balancing applications in our labs and have configured their ILO's to send traps to our centralized alarm monotoring station (ArteMon).

The ILO's are updated with firmware 2.05 and we see the test trap and the invalid logins detected alarm.

Also, whenever we shut a server down we see 2 alarms triggered at the same time:-

cpqSm2PowerOutage
smqSm2ServerReset

We were hoping that an event such as a server shutdown would trigger say the Power outage alarm, and then the event of the server coming back up would trigger the Server Reset, as we could have configured our ArteMon application to use the reset to clear the outage.

Unfortunately it seems the ILO only ever sends them together during a shutdown. Is there an explanation for why this is? Or is there a way to get it to send a clearing version of both of these alarms whenever a server is back up?

Also, is there any documentation that dictates exactly what alarms are designed to be triggered and when? And whether they are supposed to auto-clear if an event is rectified or if it is expected that they will be manually cleared at the alarm monitor station? So far we have not found anything, all we have to go on is what the mib files show up.

An investigation into the HP ILO mib files (cpqhost.mib & cpqsm2.mib) that we reigstered with our ArteMon alarm monitor revealed that the traps have only 2 bound variables, the sysName, and cpqHoTrapFlags. Looking at the mib, there doesn’t seem to be any traps that would indicate that the poweroutage is resolved. Also, the cpqHoTrapFlags have a field called TrapCondition (bits 2 to 4), with the following values:

0=Not used (for backward compatibility)
1=Condition unknown
2=Condition OK
3=Condition degraded
4=Condition failed
5-7=Reserved

Is this used or set by the ILO as a form of severity attribute?, can we configure the ILO to update these values? It seems the cpqHoTrapFlags value only ever gets set to 0, is there any utility available to provide more advanced ILO settings than its config GUI, this only really seems to let you set up to three detination ip’s and enable pass thru from agents running on the server o/s. We do actually have problems getting agents installed which we will raise in a separate support thread.

At present we are somewhat unsure of exactly what is supposed to happen and when, and if we are observing the designed behavior.

Any help and support is much appreciated,

Many thanks,

-Matt.

[davidlagace]

Does anyone work on that?
I've the same question.

I wonder how we could clear this alarm if the field is still set to 0 and not sendind a "CLEAR" notification.

[gxtoast]

Welcome to HP management. You think a company with HP's reputation would do things more solidly in this space.

Cloud in the Enterprise is coming and HP really need to get this together in order to compete in that space.

Come on HP, produce some decent management products (and then document it all properly).

[Oscar A. Perez]

Matt,

About the iLO2 sending cpqSm2PowerOutage and smqSm2ServerReset together during an OS shutdown: iLO2 can only detect power transitions on these servers. Old servers used to transition directly from ON to OFF  but, newer servers like the G6 would actually go from ON to Reset, then from Reset to OFF. 

Also, iLO2 doesn't have a way to know that the shutdown was either initiated by the user from the OS or it was an unexpected event. If you don't want these traps being sent during a scheduled maintenance, you could use the iLO2 Virtual Buttons to initiate the shutdown. This way, you tell iLO2 that the shutdown event is intentional when you press the Virtual Power OFF button on the iLO2 webGUI or send an XML script.