C220M3 - CIMC SNMP Implementation

[chrgarros]

Aloha,

I'm trying to monitor UCS hardware (mainly power usage and temperature) using the SNMP agent running on the CIMC board.

I can *sometimes* read the MIB values (.iso.3.6.1.4.1.9.9.719.1.9.14.1.2.1 for example) but then the SNMP agent hangs and cannot return any values under the private branch. (top of the three still working eg: .iso.3.6.1.2.1.1.4.0)

Looking at the CIMC log, I see that the SNMP daemon is crashing and restarting :

BMC:snmp-subagentd:7347     NET-SNMP version 5.5 AgentX subagent connected

Basically I can do one walk or one get, retrieve the value(s) and wait for the agent to restart....

I've tried to capture the packets when the agent is crashed. I see an answer generated from the CIMC IP with a "No Such Object" status. It looks like the base SNMP daemon is still running but the link to the CIMC monitored hardware broken.

I'm running firmware 1.5(1j)

Any help or feedback would be greatly appreciated.

[chrgarros]

It was a bug with the CIMC firmware. I've upgraded all my servers to the latest version and the issue is gone.

I now have a fully working cacti template to monitor and apply thresholds for the C220M3 boxes.

[Keny Perez]

Cool, thanks for the information.

I am pretty sure this may help others with the same issue!

Can you mark your own answer as the correct solution for this thread? That will give it more visibility.

-Kenny

[Jeffrey Foster]

Hi Christophe,

As you mentioned, this was addressed in a recent CIMC update - specifically CIMC version 1.5(2).

Jeff Foster

[asharwan]

Please use this link may be this will help you.

http://www.cisco.com/en/US/docs/unified_computing/ucs/c/sw/cli/config/guide/1.4.1/b_Cisco_UCS_C-Series_CLI_Configuration_Guide_141_chapter_01101.html

[chrgarros]

I've found another weird issue with the latest firmware.

In some cases, -not everytime- when a power supply is unplugged, the CIMC interface show it correctly but polling the SNMP tree report the power supply as connected :

woot# snmpwalk -v2c -c public 192.168.1.90 .1.3.6.1.4.1.9.9.719.1.15.56.1.7

.iso.3.6.1.4.1.9.9.719.1.15.56.1.7.1 = 2

.iso.3.6.1.4.1.9.9.719.1.15.56.1.7.2 = 2

..still a long road for the indian gurus before getting something working.

[Keny Perez · Accepted solution]

Christophe,

I have seen that before and what the developers have said is that we need to "use CIMC as the first monitoring tool", cause as yuo may see that logged the fault correctly.

Could you please open a Service Request so that TAC can see what is going on, explain and if necessary open a bug for this?  Otherwise, it will be hard for us to fix it, cause if the customer that faces this does not get in touch with us, probably the problem will not be heard, so a TAC case is the best option.

-Kenny