Unsolved
2 Posts
0
1502
S4810 Swtich Stack no downtime upgrade / reboot
Have a pair of S4810 switches in a stack. They are currently running v9.10(0.1P3). I uploaded v9.14(1.9P4) and the associated boot code.
I have Server racks with redundant connections to both switches for storage, etc in an ESX server environment. I want to know if I can reload the stack members one at a time to get the stack up on the new software without just reloading the entire stack.
The goal is to keep at least one up so the backend storage communications are not interrupted. I can reload one stack member. I suspect it will come back up but presume the switches running these different versions will not create a stack. I could then reload the other switch in the stack and ideally it will come back up and the stack will be whole again.
Can I do this such that one switch will always be passing traffic between switchports on a single switch so I don't loose the backend storage communications? If the stack is broken after the first switch reboot, will I still be able to reload the other member and if so, would I need console/serial access to the switch?
The alternative is to shut down all of the servers and ESX hosts but I am trying to avoid that if possible.
Thanks,
DELL-Joey C
Moderator
Moderator
•
3.1K Posts
1
March 25th, 2021 20:00
Hi,
I'm not network savvy, so I spoke to one of the co-worker of mine. The best practice is both stack switches should be be upgraded and at same time both need to be rebooted. If one switch reboot only, it would be in high availability and result to packet drops. So it's advisable to do it when you have a down time, to do upgrade on both switches.
Here's the download page for S4810 for keeps, in case if you need to check release notes.
bealdrid2
117 Posts
1
March 26th, 2021 08:00
There's not really any way to work around this with a stack. I would recommend switching to VLT which would allow you to reboot one at a time. I have heard of folks trying to "break" the stack and upgrade the units separately. Seems a bit risky to me though, and it wouldn't work if you have any cross-switch LAG connections.
Your servers may be able to ride outage and recover even if both switches go down at the same time if you were to upgrade the stack normally. But it's hard to say. You would definitely want to make sure all server OS's are configured with appropriate disk timeouts, etc to increase your odds. Worst case you'd probably be left with a bunch of disks that have gone read-only. Best of luck.
Ch3Co
2 Posts
0
April 5th, 2021 06:00
Thanks. I haven't restarted/updated the switch yet but we elected to shutdown all of the servers so the stack can be reloaded as a unit just to be safe.