3850 Member replacement in stack with slightly different model

[Seano]

Hi Guys looking for advice to tackle a replacement problem.

I have a 3850 stack 1 to 4 with different types of 3850's from SFP's to SFP + and 10GBE

I stupidly replaced the 3rd member of the stack of a 24 port SFP with a 24 port SFP+ as they were bought near the same time I though versions would be the same and in a rush did it. 

Curernt stack version looks like

Switch Ports Model              SW Version        SW Image              Mode

WS-C3850-24XS      03.07.04E         cat3k_caa-universalk9 INSTALL

and new switch looks like

Switch Ports Model              SW Version        SW Image              Mode

WS-C3850-24XS      16.3.5b           CAT3K_CAA-UNIVERSALK9 INSTALL

As you can see turning it on screwed the membership and only 2 switches were showing in a show ver.

Routing is still working from face value, and I have unplugged the new switch from the broken stack.

Im going to get them all on the same version and so on and I was going to redo all all of the new member 3's interfaces.

But can someone advise on the best way to get out my pickle?

Thanks in advance 

[Leo Laohoo · Accepted solution]

1. Disconnect the stacking cable.
2. Erase the config of the 2nd switch (the one running 16.3.5b).
3. Copy the IOS into a supported USB stick;
4. Run the software install command and downgrade the IOS (DO NOT REBOOT);
5. Power down the 2nd switch;
6. Attach the stacking cable; and
7. Power up the 2nd switch.

[Seano]

Thank you for the quick reply.

I had broken the stack properly with 2 members going into ROMMON mode and would not boot. I had to use the command 

switch: set MANUAL_BOOT no

to get it to boot. Then I got the below message with the second switch rebooting every time it did an HA sync with the master.

%ISSU-3-PEER_INCOMP_SET: Peer incompatibility set for image (CAT3K_CAA-UNIVERSALK9-M), version (03.07.04E) on peer uid (2)

<Sat Aug  4 08:40:02 2018> Message from sysmgr: Reason Code:[4] Reset Reason:Reset/Reload requested by [stack-manager]. [ISSU Incompatibility]

Through me off looking like it was incompatible with software versions when these were the original members on the same version. Cisco docs told me from the second error log to shut down the whole stack and then bring the members up one by one. I followed the instructions to downgrade the new switch and keep it off. I then booted them up in order doing a show switch showed the the stack built it self from scratch as each switch joined.