Hi,
I recently started graphing CPU busy time with SNMP and it worked fine for a few weeks, and then all of sudden one day, it always returned 0 and has done ever since. I suspect rebooting the filer would bring the SNMP CPU reading back to life, but I was wondering if anyone had a better idea...
I'm running NetApp Release 6.5.2R1P9 on an F940 and using the following command to get the information:
snmpget netapp public -m NETWORK-APPLIANCE-MIB.txt \ NETWORK-APPLIANCE-MIB::cpuBusyTimePerCent.0
Telneting to the filer and running "sysstat" returns normal (non-zero) CPU usage information.
Ideas?
Cheers, Gavin.
Gavin Kelman (2005-01-27 18:00):
I recently started graphing CPU busy time with SNMP and it worked fine for a few weeks, and then all of sudden one day, it always returned 0 and has done ever since.
Hi Gavin.
We encountered that behaviour as well, this looks like BugID 141441. Also, take look at BugID 12961 for additional enlightenment :)
Concerning monitoring in general, you may also want to check out ToasterView by Ben Rockwood: http://www.cuddletech.com/code/toasterview
I suspect rebooting the filer would bring the SNMP CPU reading back to life, ...
That worked in our case.
-Cornelius
On Thu, Jan 27, 2005 at 06:45:30PM +0100, Cornelius R. Bartke wrote:
Gavin Kelman (2005-01-27 18:00):
I recently started graphing CPU busy time with SNMP and it worked fine for a few weeks, and then all of sudden one day, it always returned 0 and has done ever since.
Hi Gavin.
We encountered that behaviour as well, this looks like BugID 141441. Also, take look at BugID 12961 for additional enlightenment :)
Ah, that explains it nicely.
Concerning monitoring in general, you may also want to check out ToasterView by Ben Rockwood: http://www.cuddletech.com/code/toasterview
It was toasterview that inspired me to setup all the monitoring. I would have used toasterview but the disks stats didn't work, so rather than fixing it, I started from scratch with RRDTool and shell script and it works very nicely. :)
I suspect rebooting the filer would bring the SNMP CPU reading back to life, ...
That worked in our case.
An upgrade to 6.5.3 looks likely, plus a rewriting of the stats to take into account bug 12961.
Thanks for your help, Gavin.
-
Cornelius R. Bartke -
Gavin Kelman