Replacing a switch in 2960 stack

[Psmurali89]

Hi All

I got 2x 2960 switches in a stack. The member switch failed

Switch 1  - Member -  switch priority is 1

Switch 2 - Master - switch priority is 5

I got spare switch, just need to check if the below steps are ok:  

* Check new switch has same IOS and hardware model. 

* Connect to console and apply "switch 1 provision ws-c2960xxxx"  <<<< I believe this line is needed? 

* remove the failed switch and connect the new switch stack and power cable and power on. 

I believe all the configs will be copied across to the new member switch? 

[Reza Sharifi]

Hi,

Have a look at this document and follow the steps to add a new member to the stack.

https://www.cisco.com/c/en/us/td/docs/switches/lan/catalyst2960x/software/15-0_2_EX/stack_manager/configuration_guide/b_stck_152ex_2960-x_cg/b_stck_152ex_2960-x_cg_chapter_010.html#con_1178943

https://aboutnetworks.net/how-to-add-a-switch-to-an-existing-c2960x-stack-without-breaking-everything/

HTH

[Psmurali89]

The second link was helpful, thank you. 

To confirm, I just checked the switch and the priorities are left as default (1) and the master was elected based on lowest MAC address. Am going to change the priority of master to 15 and member to 10. 

Out of curiosity, if master fails then the other switch with priority of 10 becomes master and when I add new switch in, the new switch priority will be 1 (default) and it stays this way until I change it and this wont cause any issues?  

[Reza Sharifi]

Hi,

Am going to change the priority of master to 15 and member to 10. 

Yes, I recommend giving each switch a priority number. This way, you always know what switch is the master and what switch is the member. Once you make this change and all working, I also recommend physically marking each switch as "master" and "member"  this way, if one of them fails, you know which one was it. 

Out of curiosity, if master fails then the other switch with priority of 10 becomes master and when I add new switch in, the new switch priority will be 1 (default) and it stays this way until I change it and this wont cause any issues?  

That is correct. In your example, if 15 fails, 10 will become the master and once you replace it, you want to give the new switch a lower priority than 10 (5, 6, etc..) this way there is no change, and no reelection. 

HTH

[Psmurali89]

Thank you very much. 

 I believe changing priority will take effect immediately and doesn't require reboot? 

[Reza Sharifi]

New priorities will not take effect until you reboot the stack.

Have a look at table-2 in this link:

https://www.cisco.com/c/en/us/td/docs/switches/lan/catalyst9200/software/release/16-10/configuration_guide/stck_mgr_ha/b_1610_stck_mgr_ha_9200_cg/managing_switch_stacks.html#concept_j2w_shc_31b

HTH

[Psmurali89]

Sorry Am bit confused, So if this doesn't take effect until rebooted then for example member switch fails (I have set the priority to 10 on member and 15 in master beforehand) then after adding new switch it has default priority of 1 but the master switch priority of 15 would not have taken effect as the master switch has not been rebooted so the priority will be still 1 and the election may take place in case if the new switch added has lower mac address?

[Reza Sharifi]

So if this doesn't take effect until rebooted then for example member switch fails (I have set the priority to 10 on member and 15 in master beforehand) then after adding new switch it has default priority of 1 but the master switch priority of 15 would not have taken effect as the master switch has not been rebooted so the priority will be still 1

That is correct. Every time you change priority, it will not take effect until you reboot the switch or the stack. The only time the stack goes through reelection is when the stack is rebooted or when the master fails, which triggers an election and the switch with the highest priority will take over.

[Reza Sharifi]

Also, please make sure that when you make this type of change, you do it in a maintenance window as sometimes you never know what can go wrong. 

[Psmurali89]

yea sure, thanks. I also need to change the PDU where these 2 switches connected. The below summary is based on No switch failed and if a switch failed while changing the PDU. 

To summarise,

* Change the priority value of master to 15 and member to 10 --- No reboot required at this stage as suggested in the link you provided.

* power down member switch and connect to new PDU and power ON - if this switch failed to come back then replace with spare switch that has same IOS and hardware version. Once connected to stack cable and powered up, the config will be synced from master (priority value of 15) and the new member switch will have a priority of 1. 

*If the member switch has been replaced then change the member switch priority back to 10. 

* power down master switch and connect to new PDU, at this stage the member switch (priority 10) will become master and once the master switch comes back online it will become master as before. If the master switch fails to come back then replace with spare switch which has default priority of 1 and become member and will stay as it is. 

Does the above sounds OK? 

[Reza Sharifi]

The above steps should work, but if you will have a maintenance window and can afford to turn off both switches, it is easier to:

* Change the priority value of master to 15 and member to 10. Save config

* Turn off both switches, and then connect them to the new PDU.

* Turn on the master and right after that turn on the member.

* After the switches boot, an election should take place and the switch with priority 15 should be the master and priority 10 member.

This is assuming that the stack cables are connected, both switches are running the same exact IOS version, and have been provisioned correctly. 

HTH

[Psmurali89]

Thank you. 

I really wish to do this way but unfortunately we cant afford full downtime. The HA firewalls and servers with 2 NIC's are connected in this stack. For Example, Firewall 1 is connected in switch1 and firewall 2 is connected to switch 2. So taking down one switch at a time will minimize the downtime. 

I will try to do as per the steps I mentioned in previous reply. If I can get full downtime (still pending confirmation) I will follow the steps you mentioned above. 

[Joseph W. Doherty]

Doing one switch, at a time, as you originally planned, should work pretty much as you desire.  Basically, your conversion is mimicking two times when a switch member fails.

I.e. I don't believe you need to schedule both switches being down, unless you really, really believe you need to manage which switch is the active master (again, not something I've seen which really makes much of any operational impact).

That said, still a good idea to do this change during minimum business usage.

[Psmurali89]

Sure thank you. 

[Reza Sharifi]

You should be fine. Just have a plan B ready in case something goes wrong, especially if you are doing this late at night or after midnight because depending on where you are getting TAC help at that time sometimes is not easy. I am not trying to scare you but rather prepare you because I have seen quite a few cases where when you make a change to a stack, the whole stack will have a heart attack and nothing works until you reboot the whole thing. Also, as part of your prep work, I also recommend you open a ticket with TAC ahead of your maintenance window. This way, you already have a case number ready to go just in case.

Good Luck!