Replace stack base unit on live network

[nodeisup]

I have a stack with 5 switches and one of them is base switch. Currently we have to replace this switch regarding a fan issue. For replacing this switch I have following questions:

-Can a new Stack cabaling make an issue for the stack on live system?
-Does AUR function work also for base switch? How could be saved configuration on base switch after installation?
-Must the stack rebooted after replacing?

Thanks in advance.

[Flintstone]

Hi nodeisup,

When you remove the base switch there is a backup base switch which will take over the operation.  I would then copy the config from the old base switch to the new base switch, so when you add in the new base switch, then there shouldn't be any issues?

Additional stack cabling shouldn't be an issue on a live system until you plug it into a new stack switch?

I would personally do all this during a change window out of hours in case anything goes wrong.

CheerZ and good luck

[shezzski]

1.Can a new Stack cabaling make an issue for the stack on live system?
2.Does AUR function work also for base switch? How could be saved configuration on base switch after installation?
3.Must the stack rebooted after replacing?

1. If you have redundant stack cabling ie a full ring then the process of removing/replacing a unit and connecting up the stacking cables will not cause an outage.  Where you run into problems is if you have a stack of 5 units with no return cable to the top of the stack ie not redundant (you can check with "show stack health") and you say remove unit #3 in the middle of the stack.  This will separate the stack and depending on how its connected to the rest of the network you can run into all sorts of problems.

2. Well AUR works well on later codes, as it incorporates DAUR, AAUR, LAUR and regular AUR.  In theroy unit #2 is supposed to hold a config backup for the base unit so that if it fails unit #2 can copy the config back to the base unit.  This should not be an issue on a stack of 5 units.  Once again you will run into problems when you have a stack of 2 units and are trying to replace the base unit.

3. The stack is not required to be rebooted once you replace a failed unit, though this is the base unit, im not 100% sure if the stack will reset itself automatically.  In the later codes if you replaced a failed stack member the stack will upgrade the diagnostics, reboot the failed switch, upgrade the image, reboot the failed switch, download the config to the failed switch and then reboot the failed switch so that it is back "operational" and "healthy".

[CptnBlues63]

I have a stack with 5 switches and one of them is base switch. Currently we have to replace this switch regarding a fan issue. For replacing this switch I have following questions:

-Can a new Stack cabaling make an issue for the stack on live system?
-Does AUR function work also for base switch? How could be saved configuration on base switch after installation?
-Must the stack rebooted after replacing?

Thanks in advance.

I've labbed quite a few scenarios with a downed switch in a stack including the base unit.  What I discovered was the base designation does not come back to the base unit without rebooting the entire stack.  If it were me, here are the steps I'd follow:

1) Ensure replacement switch has same FW/SW as stack members
2) Backup config of existing stack
3) Apply base unit config to replacement
4) Set "base" switch on replacement unit to the "on" position.
5) Power stack down
6) Replace base unit, make all connections (including cascade cables)
7) Power stack up and ensure everything is working as it should be and configuration is correct

If you can't reboot the stack and have to replace the base live, follow the above omitting steps 5 & 7.  Keep in mind though in this situation that the base designation will stay with whichever switch picked it up temporarily until you next reboot the stack.

The stack will not reboot itself automatically.

[Johan Witters]

Hi,

you can perform the following check:
1) check the stack cabling with "show stack health": it should say "Resilient". If not, check your cabling
2) save the config on-box with "wr mem"
3) take an off-box copy of the config with "copy run tftp address x.x.x.x filename backup_stack.cfg" where x.x.x.x is the ip of a machine with a tftp client installed and running (eg tftpd32)
4) check auto-unit-replacement with "show stack auto-unit-replacement", it should indicate all configs are saved
5) unplug the base unit, unit 2 should become temp base
6) install the new switch, but leave the base switch on the back to the left (non-base unit)
7) link all stacking cables and boot the switch
as soon as the switch is up and running, verify the stack with "show stack health"
9) If the stack is fully formed, move the "base" selector on the back of unit 1 to base and reboot the stack
10) verify correct configuration of the stack after boot.

If you you have the luxury of downtime do the following.
1) take an off-box copy of the config with "copy run tftp address x.x.x.x filename backup_stack.cfg" where x.x.x.x is the ip of a machine with a tftp client installed and running (eg tftpd32)
2) reset the stack with "restore factory -y", the stack reboots
3) during boot, unplug the power cables
4) replace the base unit and boot the stack
5) put the backup back in