In my case, a remote OC300 failed firmware upgrade via cloud last night and is stuck at 90% with exact same error. I have tried rebooting twice (remotely using a ethernet power controller) and situation has not improved. Unfortunately, we dont have a backup of the configuration to restore. The APs are mounted high up in warehouse. In this situation, while everything is working for now, without of course any visibility, what happens if we simply replace with a new controller and configure it for almost identical configuration (device credentials will of course be different as each controller has a random password) in terms of interfaces / vlans / IPs, SSIDs / passwords? How will the existing APs and switch / gateway enroll in this situation with old controller having pushed the device password to the all these devices? Do I need to physically access all devices and do a factory reset on them?

Thanks

@biomed32uk 

It's not good is it, the update needs to be revoked while it's semi bricking peoples controllers.

There is obviously something in certain configs that is causing the update to fail, as a clean install it is fine but I'm not wanting to start from scratch as my switchports, router, vpn, port forwards etc are all in that config.

Perhaps the next hardware version of the controller needs to be more like some switches where they are dual firmware capable with config etc, so it's an easy robust rollback in case of trouble.

I think you can still log into access points remotely to reset them so you can adopt them back into another controller. Not sure what happens when they are rebooted and the controller isn't there. I presume in that state they are then useless as there is nothing to tell them what to do.

Definately needs to be a bit more robustness in this system, one for tplink to take note of !.

Hi, 

The same here. I have provided 12 different installations and environments with the new update. All environments that did not have an SDN-compatible switch in the system ran flawlessly. Three environments with switches managed by Omada ran into the same issue during the update and had to be restored via recovery mode, downgrade and backup recovery.

I suspect a relation here as soon as SDN switches are involved. At least that's what it looks like at first glance. For a customer with an SDN-compatible switch, which was in pending mode, the update also ran cleanly. Only as soon as switches were integrated into Omada, the update no longer ran.

I use many different SDN compatible switches, e.g. T1500G-10PS v2.0, TL-SG2428P v1.0, TL-SG2008P v1.0...

Maybe this will help with troubleshooting.

Best,

Ronald

@bincomIT 

I have experienced the same thing, do you have switch ACL on your installations? I have found that if I disable the switch ACL then there are less problems.

@shberge 

I have environments with switch ACL and without. The three environments that could not be upgraded have the only similarity of having an SDN managed switch available. All other environments do not and the update worked cleanly there.

In my case, one site with SG2210P switch, OC300, ER7206, 6xEAP225 went fine from 4.4.6 to 5.0.29 all thru remote cloud software upgrade. Then I proceeded to do the same for another site with exact same set of equipment and also running 4.4.6. And that controller never came back up. I wil plan to go onsite next week and try my luck doing firmware recovery. If 5.0.29 still does not work, I will try with firmware 4.4.6. Customer lost the last backup as they upgraded OS and re-purposed that PC for something else. Hopefully I will not need to do factory reset else I will also need to factory reset all APs, switch and gateway since we did not keep track of the device password either for this site.

And I don't have any switch or EAP or Gateway ACLs and I still have controller stuck at 90% even after two reboots.

@dpsguard 

I recommend taking a small USB stick with you and setting up the automatic backup. Then you always have the backup on the controller.

You may also be lucky, because in two of my environments where I had to execute a recovery and downgrade to 4.4.6, the last config was still active without having to restore a backup.

@bincomIT 

I believe automatic backup only works if OC200 is powered from POE, correct?

@supermarkert 

Not 100% sure, as I operate all controllers exclusively via PoE. But at least the manual says so.