C9300 Stack OS Upgrade non-downtime

[inb]

Hello.

I am running a stack of 2 C9300s.
The OS is using 16.12.14, which will be a problem when upgrading the OS later.
I know that all switches are rebooted when upgrading in the stack state.

Switching
I thought of the following as a way to minimize downtime.

1. sw 2 all ports shutdown
2. Remove data stack cable 1
3. Remove data stack cable 2
4. sw 2 os upgrade progress
5. sw 1 all ports shutdown + sw2 all ports up (downtime max 60sec-STP calculation)
6. sw 1 os upgrade progress
7. sw 1 all ports up + sw2 all ports shutdown (downtime max 60sec-STP calculation)
8. Connect data stack cable 1 (connect sw 2 stack and reboot)
9. Data stack cable 2 connection
11. Complete after booting sw 2.

Is there a better way other than this?
Please help.

[balaji.bandi]

I do not believe - any other process or aware of as of now  with out rebooting whole stack will have new IOS effective

*in the Future may have option hot upgrade i am hearing. not yet ready for these models.

if you do 1 at a time, you have version mismatch, they can not form the stack.

best is always upgrade who stack and reload with the expected downtime.

If your environment required very high availability need to consider different model of deployment stackwise virtual (with the end device has dual home connectivity)

[inb]

Thank you for answer.
When step 6 is finished, both sw 1 and 2 operate in the same os.
If you connect the data stack cable in step 8, it is expected that there will be no problem.
This switch is in a very important position during initial configuration
It was installed without reviewing the part for os upgrade.

[Leo Laohoo]

Boy, do I have a surprise for you!

The stack is currently running 16.12.X.  Have a look at the options available for "reload ?".  Do you see the option for "reload fast"?  
That option is good if you're upgrading from the same train, i.  e.  16.12.X to another 16.12.X.  

The next version is 17.X.X and upgrading to this version will not be easy because there are be extra steps in addition to the "install mode".  And when upgrading the ROMMON, it will require a mandatory reboot. 

[inb]

Thank you for answer.
Does it have anything to do with the reload command?
Currently operating in install mode.
I know the command for reload slot x.

[balaji.bandi · Accepted solution]

I would always recomend to reboot whole stack for best outcome, until you want to try some experiment in live environment not having sleepless night to run to work to fix the unknown issues.

[paul driver]

Hello
FYI for a fast reload one caveat I am aware of is that it isn’t compatible when you have port/link aggregation enabled within the stack and seeing as you have just a stack of two switches and jumping to a higher ios code then probably a ISSU upgrade would be more beneficial as this should provide the shortest downtime for you network and it will basically perform the same function as you have describe in your OP, in that the secondary (standby) switch will be upgraded first and then reloads and when this comes back on-line it will resume the master role and then the previous master will be upgraded and reloaded and as/when that comes back on line will resume it original master role, however you’ll need to check if this feature is supported for the 93K, if it isnt then probably you should go for the basic process of the "Install add file install <tftp -usb> xxxx.bin flash: and request platform software package install "… and automate or schedule the reload of the whole stack.

[Audrius Jonavicius]

Hi,

I think that now it is possible. Take a look here https://community.cisco.com/t5/networking-blogs/xfsu-on-catalyst-9300-the-always-on-access-network/ba-p/4273159

Regards,

Audrius