3850 stack elevated CPU

[richard.schultz1]

Hey all,

I have a 3850 stack with consistently elevated CPU (right around 60%). The stack serves as a middle man between an ASA pair and a pair of 6807 switches in VSS mode. L2 and L3 (behind the firewalls) are all concentrated on the 6807's - this stack isn't root for anything, nor does it have any routes. However, I'm seeing the following in the attached text document. This is my first experience with 3850s - is this a normal behavior given the topography location of the stack? We had a similar setup with 3750s and 6509s (obviously not VSS) and the 3750s were running cool.

ASA (A/S HA) -> 3850 stack (VSS) -> 6807 pair (VSS) -> multiple 3850 stacks.

Stack information:

Switch Ports Model SW Version SW Image Mode
------ ----- ----- ---------- ---------- ----
1 56 WS-C3850-48T 03.02.03.SE cat3k_caa-universalk9 INSTALL
2 56 WS-C3850-48T 03.02.03.SE cat3k_caa-universalk9 INSTALL

[Reza Sharifi]

Hi Richard,

CPU running at 60% on the 3850 is not normal.  We have several 3850s with the same version of IOS you have and had a lot of traceback messages which would cause the switch to crash and reboot.   After consulting with TAC, they recommended "03.06.03.E".  This version seems to be running fine with no traceback messages and the CPU runs around 20%.

HTH

[richard.schultz1]

Reza,

I do have a CRQ in place to upgrade IOS to that exact version to fix other bugs I've noticed pertaining to hardware health monitoring (switch says the temps are good, SolarWinds says they're high).

Unfortunately this means downtime which is as you know Management bait for panic... Hehe.

I'll keep an eye on the stack and try pushing the CRQ through earlier.

[Reza Sharifi]

Richard,

Unfortunately this means downtime which is as you know Management bait for panic... Hehe.

I completely understand that.  I deal with the same panic issue all the time :-)

We have also SolarWinds and see the same exact issue regarding what the switch say vs the monitoring device.

Anyway, I would upgrade sooner than later, specialty if you start seeing traceback messages, if not you will get an alert/call at 2:00AM.

Also, 03.07.02.E seems to be running ok as well.

HTH