Re: [Observium] Discovery crashing UPS management card
For the record, Eaton have confirmed that they have reproduced the fault
on the 9xx range of UPSes including on the latest revision of management card so it is being raised as a bug and is not just an issue with my particular hardware. That's good, I guess (hope) they'll release a fixed version...
We can't really get around the table reading. What you could do in the meanwhile is add UPS-MIB to the $config['os']['powerware']['mibs'] array, and remove XUPS-MIB from there. Note that it'll affect all your powerware UPSes... but I guess that should stop it from crashing. Barring any bugs in their UPS-MIB table walk code *G*.
Ideal situation would be to not crash on a simple table walk of course ;-)
Tom
Just upgraded to this week's latest community release, and it is presumably now not walking the tree, as it is no longer triggering this crash, and is now picking up the other voltages correctly. Great job on fixing a bug that wasn't yours to fix before I'd even told you about it :-) Hopefully Eaton will be equally efficient!
Am very impressed with what I've seen so far in Observium in a week's dabbling. Am certainly contemplating taking out a subscription to stay current once I've done some more testing and added the rest of our equipment in, and ideally found some way to make it talk to the equipment that doesn't do SNMP...
Steve.
On 04/03/2014 12:12 PM, Steve Kersley wrote:
on the 9xx range of UPSes including on the latest revision of management card so it is being raised as a bug and is not just an issue with my particular hardware. That's good, I guess (hope) they'll release a fixed version...
We can't really get around the table reading. What you could do in the meanwhile is add UPS-MIB to the $config['os']['powerware']['mibs'] array, and remove XUPS-MIB from there. Note that it'll affect all your powerware UPSes... but I guess that should stop it from crashing. Barring any bugs in their UPS-MIB table walk code *G*.
Ideal situation would be to not crash on a simple table walk of course ;-)
Tom
Just upgraded to this week's latest community release, and it is presumably now not walking the tree, as it is no longer triggering this crash, and is now picking up the other voltages correctly. Great job on fixing a bug that wasn't yours to fix before I'd even told you about it :-) Hopefully Eaton will be equally efficient!
Am very impressed with what I've seen so far in Observium in a week's dabbling. Am certainly contemplating taking out a subscription to stay current once I've done some more testing and added the rest of our equipment in, and ideally found some way to make it talk to the equipment that doesn't do SNMP...
Hah. Very nice. I rewrote a lot of the sensor code recently (as in, after previous CE), I guess we approach it somewhat differently now. We rock ;-)
We don't have any plans to support non-SNMP speaking devices (well, I would like to, but it would require a major overhaul of poller, discovery and parts of the web interface - doesn't go very well with the fully-automatic-everyting approach); I don't have many that don't speak SNMP - even my tape drives, label printers, etc speak SNMP... Unfortunately my projectors only do traps when overheating and such, I can't realtime monitor them :D
Tom
participants (2)
-
Steve Kersley -
Tom Laermans