Forum Discussion

smp_86112's avatar
smp_86112
Icon for Cirrostratus rankCirrostratus
Jan 05, 2011

No Log Msg Removing PSU?

For demo purposes I pulled one of the power supplies from a 6900 LTM, and was surprised when I received no trap or log message. Pulling the cord and flipping the switch do result in traps and logs. The only reason I can think of is that F5 decided no one needs to be notified when intentionally removing the PSU, but I have a hard time swallowing that idea. Anyone else have any thoughts on why this might be?

 

 

6 Replies

  • Interesting. I've gotten traps/logs when losing a UPS to which my PS is attached so I don't get why switching it off wouldn't trigger something.
  • Hi Chris, thanks for looking.

     

     

    Just to be clear, switching it off and on DID result in an alert. Pulling the power cord DID result in an alert. Pulling out a fully functional power supply DID NOT result in an alert (or any log messages for that matter). Given those three scenarios, what you described seems to match what I described as "pulling the power cord", which in my case DID result in an alert. I yanked out a running power supply which DID NOT alert. I think those two scenarios are different because in your case the power supply bay was populated, and in my case it was not.
  • I'd be fine if yanking a power supply that was turned off/unplugged didn't alert. But yanking a working one should be the equivalent of switching it off which should definitely alert. If no one replies here, I should think a support case/CR is necessary.
  • there's an oid that you can poll (/usr/share/snmp/mibs/F5-BIGIP-SYSTEM-MIB.txt)

    sysChassisPowerSupplyStatus OBJECT-TYPE
            SYNTAX INTEGER {
                    bad(0),
                    good(1),
                    notpresent(2)
            }
    
  • Thanks Jason. I guess I'm surprised that if I can poll, then why not send a trap?

     

     

    Looking at /etc/alertd/bigip_system_check_error_maps.h, it seems there are a number of events related to a power supply that will result in a trap. But none seem specific corresponding OID. Is there a way to force the LTM to generate a trap when the OID value is bad (0) or notpresent (0)? This area between configuring traps and the alertd-maps files have always been a bit murky to me, as I haven't found the doc to be very clear.
  • Assuming you were getting the message in a log file, you could create a custom snmp trap as described in Solution 3727 (Click Here) I'll see if I can track down what logging subsystem power messages are in and what level is required to see the ps removal messages.