Passport 8600 CPU Swap Command

[riscyrich]

Anyone know what the CPU swap command is for PP8600?

Many thanks

Rich

 

[curtismo]

reset -y

 

[riscyrich]

Isn't that the command to reset the whole switch and not swap the cpu's?

If we reset the switch then it will just go back to using the CPU in slot 5...

 

[curtismo]

If you're running in the usual Master/Slave configuration, performing this on the current master CPU will cause the other CPU to boot, reload the R/RS card software and become the master.

I believe there is a boot flag that sets whether CPU slot 5 or CPU slot 6 is the master upon cold boot.

Since I don't run mine in HA configuration, I am not sure how to do this in that situation.

 

[curtismo]

You can also press the recessed "reset" button on the front of the master CPU.

 

[DaddyOfThree]

You can also issue the following command (does the exact same thing that curtismo has provided);

config sys set action cpuswitchover

Cheers!

 

[riscyrich]

Thanks guys, now to find out why the CPU card keeps crashing :'-(

 

[curtismo]

CPU Crashing? Can you provide more information? What version of 8600 code do you have? Are you running anything like ISTs or SMLTs?

Logfile output is also helpful.

 

[riscyrich]

Morning guys,

Yeah, the active CPU card just locks up. Can't get anything from the card at all on any of the ports, management, serial, modem, SNMP etc. Won't respond to ping, can't make any routing decisions, become a VRRP master etc. However the strange thing is the PP keeps on switching traffic. I’ll admit there are only a few devices plugged in as this PP is at our backup facility. It no longer has any MLT’s, SMLT’s etc. It has one connection (fibre) on to our WAN fibre ring, one connection (copper 10/100) to a Baystack 460 (stack of 4 units) and one connection (fibre) to a BES-120, nothing that’s really taxing to be honest.

We’ve swapped over to the standby CPU and are monitoring now, it was about 4 months since the last crash so we’re just waiting to see now…

Current code is: 3.7.17.0

Both our PP’s are running the same code and the main one has been fine….

Suspect it could be the card itself ? ?

Regards,

Rich

 

[curtismo]

If you were having regular crashes and have swapped to the other CPU and they stopped, to me the CPU card would be suspect.

 

[anthonyanderberg]

I'd agree that the module itself is in question.

The 8600s use a distributed forwarding model so (as I understand it) all of the line cards make switching decisions on their own, the CPUs compile and distribute routing/switching tables to each line card. This means that as long as nothing changes they can continue to pass traffic even though their CPU is locked up.

 

[riscyrich]

Ah, makes sense now...

No crash so far, wondering if there's any mileage in sending the suspect card off for testing in isolation?

My concern is they share switching fabric which could mask the problem if the fault were to lie in that area. I suppose I could pull the suspsect card if it does crash again to rule out the switching fabric.

Sadly the process of elimination is not a short path to resolution :-(


Rich

 

[curtismo]

If you have maintenance on the 8600, RMA the blade.

In the past I have swapped out a 10-slot chassis once when I was having problems with routing between the top and bottom halves, but that was long ago.