I have an issue with VSS setup. The VSS setup act as a transit core switch for internet traffic and the problem arises when there are about 600 Mbps of traffic passing through the switch. We notice that there are a lot of packet being software switch and the rate is quite high, almost like 100k to 200k per second.
The VSS Switch runs ospf and BGP and have full internet routing. Upon investigation of high cpu utilization, i notice that the switch TCAM is full and tried to set the mls cef max route to maximum value and rebooted the VSS.
Unfortunately, the problem persist after reboot.
We planned to filter full internet routes and allow default route and ospf route only in the VSS and reboot the VSS to ensure the TCAM table is not overloaded.
If that fails the next step is to revert to standalone mode and hope the cpu wont shoot up.
My Question: Will these step prevent the issue? Any suggestions or Reference similar to the problem i am facing?
unless you have PFC 3CXL and all DFCs 3CXL the TCAM can handle only 256,000 IP prefixes and nowdays a full table is in the order or 310-320,000 routes.
see table 1 in second link it tells what I have written in the first line.
>> We planned to filter full internet routes and allow default route and ospf route only in the VSS and reboot the VSS to ensure the TCAM table is not overloaded.
You shouldn't need to reload the device after accepting only default route in OSPF or BGP TCAM usage will be greatly reduced it is just question to wait few minutes (unless there is an underlying SW bug) to have the MSFC to update the PFC and then all the DFCs
Hope to help