12-17-2015 11:29 AM
What are the effects of enabling SNMP debugging on a production switch in terms of performance impact? I've been wary of doing it for fear of the switch becoming inoperable.
Solved! Go to Solution.
12-17-2015 11:59 PM
as long as you dont have multiple syslog servers defined, in general you should be ok to turn on whatever debug you like in most cases. Just check the CPU load after you have enabled all your debug and try to use a more narrow debug mask like "logging level debugging system process snmp" etc.
As an asside, I have seen one case where user+security debug logs were turned on plus there were 4 syslog servers defined. This drove the syslog process CPU high and caused some impact to ipsec setup times.
Having said that, the SNMP debug is not particularly useful on the controller for troubleshooting connectivity issues etc. rather its more debugging about how the various applications populate the mibs during a snmp request.
12-18-2015 05:58 AM
Thanks for that info, Jeff. Very helpful.
The reason I asked the question is that we're seeing a wierd behavior from certain of our Mobility Access switch stacks - the switches stop responding to SNMP 'get' requests at a certain time each day, for a predictable duration, on specific switches only. At the end of the 'timeout period', they resume normal SNMP communication. I have TAC cases open with the MAS group and the AirWave group trying to figure it out. The tech from the switching group asked me to run the 'logging level debugging system process snmp subcat all' command on a stack located in our high school's MDF and I got a little spooked about doing it, at least during regular school hours, because I've heard stories of devices becoming inoperable when in debug mode.