Recover AP250 after firmware update failure - Can I do w/o removing AP? Do I use TFTP?

[mlarma]

I have nearly 200 of the AP250 access points, but one failed an update and I am pretty sure it boots into a bootloader to the like. The reason I say that is:

- I'm remote (I'm in the US, it is in Ireland)
- I can toggle the port and reboot it. Left it down overnight once as well.
- It grabs an address from DHCP
- It shows up in the ARP cache and MAC address table
- The full stack isn't there. I can't ping it, SSH and HTTPS aren't responding.

I've done recovery on plenty of other devices where it will want to pull down a firmware file from a TFTP server. I know that I can specify settings in the DHCP options on the DHCP server so that it knows where the TFTP server is and even a boot file, but I don't know if the AP250 even supports that sort of thing. Is there some guide that I can follow to recover this guy without having someone on-site search for a tall enough ladder and pull it down and either swap it (with one we'd have to ship) or use the console connection? Any special port configuration needed on the switch?

Thank you very much in advance.

[w1f1n00b · Accepted solution]

Hopefully someone else can chime in with something more helpful, but figured I'd share my experience. I've had around a 4% failure rate on my AP250s ( over 10% on my AP550s ). I've seen them fail in a variety of ways, some similar to what you're describing. Often while showing offline in XIQ it will have a white LED on the hardware, and even still be broadcasting SSIDs but blackholeing client traffic. In almost all cases an AP in this state will not recover by any means and must be RMAed. I've had almost no luck with TFTP flashing to recover failing devices. When consoling in, it will usually hang during boot process.

[mlarma]

I was able to determine from a neighboring AP that it was broadcasting SSIDs from the prior config (not its prior config, but the enterprise).  We turned turned on the captive portal on the guest network and that registration SSID isn't showing up.  That seems to align exactly with what you're describing.  We have a little over 190 deployed and thankfully failure rates have been very low so far.  Am wondering if a reboot prior to an upgrade would help.  That wouldn't surprise me if it did.  I greatly appreciate your response above as it matches our situation 100% and confirms what we thought the path forward would be.