topic Re: Update firmware Multi EAPS ring in ExtremeSwitching (EXOS/Switch Engine)

Update firmware Multi EAPS ring

JohanHendrikx — Tue, 25 Oct 2022 06:26:39 GMT

We have a multi EAPS ring. What is the best practise how to upgrade the swtiches to prevent to create a loop?

Re: Update firmware Multi EAPS ring

dpanev — Fri, 28 Oct 2022 12:07:54 GMT

Master node first and then transit node for each ring. So the master will remain the master after the reboots

Re: Update firmware Multi EAPS ring

JohanHendrikx — Mon, 31 Oct 2022 06:58:43 GMT

How do I treat the partner and controller switches of the shared-port?

Re: Update firmware Multi EAPS ring

dpanev — Mon, 31 Oct 2022 08:13:47 GMT

Example :

1. Update Master Node

2. Reboot Master Node (Wait till eaps ring is up again)

3. Update Partner Node

4 Reboot Partner Node (The Old master should now be the current Master again)

5. Update and Reboot all Transit Nodes (This can happen in one step and you dont have do do it one by one)

Re: Update firmware Multi EAPS ring

JohanHendrikx — Thu, 03 Nov 2022 12:37:57 GMT

I'm a little bit confused. We have two EAPS rings and a shared link configuration. So in every ring I have a master and transits nodes and for the shared link ther is a controller (is an eaps transite node) and a partner node (is an eaps transite node).

So what is then the best order to reboot those switches.

Re: Update firmware Multi EAPS ring

FredrikB-NN — Fri, 04 Nov 2022 14:44:21 GMT

Ok, so the shared port(s) are not connected to the EAPS master(s)? If you upgrade within a major EXOS release, I believe you should be fine (within EXOS 30.x for instance), but bigger hops may be troublesome. I would get some advice from the TAC in that case. As a baseline, upgrade everything to the latest in the master release (latest 22.x if you're on EXOS 22) and then upgrade to the latest in the next step (30.x in that case) and so on until you reach the desired release. Landing on the latest in each subsequent major release will give you maximum feature parity with the next major release.

If you want to play it safe, disable ports so no loops can occur when a switch is rebooted. You will lack redundancy anyway during the reboot. Do the disable port in an adjacent switch, not in the switch being rebooted. Also disable a strategic port in the second ring as well while rebooting the switches in the first one and vice versa. That will prevent any strange mishaps.

On a general note, I often like to reboot the units before an upgrade if the units have been up for quite some time. That gives you the most stable state in terms of memory leaks and other bugs.