m4300 firmware upgrade without killing entire network ?

[Olivier421]

Hello Netgear community,

I thought M4300 stacking (2+ switches) would provide a way to do iterative upgrades, eg: not reboot the entire stack in one go.

Of course, this is relevant only when you have NIC bonding/lacp/lag of 1 server/appliance/switch connected to the 2 switches of your stack.

The obvious reason is not to kill the entire network for the "2 minutes"  switch restart, when doing a firmware upgrade.

I have raised ticket to support (#45297664) because of an upgrade issue, and by 2 times i got confirmation that this is "not supported". Restarting all switches in same time for firmware upgrade is a MUST.

Is that for real ?

How do you workaround this ?

I can't imagine bringing down core network switches for doing once/twice a year for firmware upgrade (because of vulnerabilities found and/or bugfix).

Thanks for your help.

[LaurentMa]

Hi @Olivier421 ,

You can follow this procedure for upgrading an M4300 stack without rebooting the whole stack - but rather rebooting the switches one by one.

Since this is your core, it would be wise if you could "practice" using another M4300 stack, but on our side we have tested this numerous times and this works.

1. Upgrade the whole stack with new version (for instance, download new firmware in IMAGE 1, provided IMAGE 1 is the next image to be used, and the setting auto-upgrade is ON for Stack Firmware Synchronization)
2. Reboot the Member units one by one (go to Maintenance / Reset / Reset unit xxx only) and wait until it's done each time
3. Reboot the Redundant Backup Management unit (go to Maintenance / Reset / Reset unit yyy only, with yyy being the backup management unit) and wait until it's down
4. Reboot the Management unit (go to Maintenance / Reset / Reset unit zzz only, with zzz being the management unit) and wait until it's down

At this stage, the entire stack runs the new firmware. But your hosts (connected using LACP LAGs across two or more switches) shouldn't have been disconnected, ever.

[Olivier421]

Hello Laurent,

that's exactly what i did (cause i was your other thread telling the same 🙂 ) on a stack of 2x m430012x12f (non vital ones) and the secondary switch was always restarting on old firmware (even setting Next image to new firmware).

The same happened from 12.0.4.8 to 12.0.15.4 , or from 12.0.4.8 to 12.0.8.15 (intermediate old one)

When upgrading node 2 to v12.0.8.15 , I saw the 2nd switch restarted with proper firmware 12.0.8.15 and 5mins after it rebooted again without reason, and went back to old firmware again. I have screenshots showing that.

All the story is within my ticket #45297664 with screenshots.

So far, I did not find any upgrade path guidance for firmwares on M4300 (going from 12.0.4.8 to 12.0.15.4) , so I apply latest directly

If what you tell is True, which i believe it is 🙂 , it was maybe not true for v12.0.4.8 or there was a bug with stack upgrade.

[LaurentMa]

Hi @Olivier421 ,

OK, sorry about that. Please give us some time, so that we can retest and address any issue on our side. I will try to have this re-tested and verified today. I will come back on this by (maximum), tomorrow.