[REQUEST] Unifi AP is constantly rebooting/restarting

phaZed

Well-Known Member
Reaction score
2,961
Location
Richmond, VA
I'm hoping someone can help me here.. I'm hoping it's not time to blow away the controller software(Win Server 2012R2). I have backups if need be.

I have swapped out the AP itself and the POE injector for a known good unit, with the same effect.

This is a UAP model from 4-5 years ago and has been working for that entire time.

It is hardwired but is wireless isolated from other APs - no wireless uplink; building construction prevents it.
It connects to the same physical switch as 5 other AP's that are working without issue.

What it does:
The unit will power on and blink yellow/orange for about 1-2 minutes then will connect to the Unifi controller. During this 2 minutes all WIFI is down.

During the 1.5 minutes it is up, the internet works, clients can connect, and the controller sees it and I even had it do a firmware update (albeit it took a few tries due to the restarting).
After 1min 35seconds to approx 2 minutes the unit will disconnect and start the yellow "connecting" light all over again.

At the point of restart, the network switch's 'link light' goes out - signaling to me that the unit is truly doing a full power reset.

I have "Forget this AP" and adopted it again numerous times, even giving it a new Alias in case the controller had an issue with ID's/Database or something like that. Same issue. Adoption works fine.
It's important to note that the AP seems to stay on indefinitely when it has not been adopted yet (at least 10-15 minutes; longest I waited).

Bringing a "new" AP onsite with "new" POE injector produces the same result which is what has me stumped. So it would seem the hardware is working?

To add to the confusion, the "bad" AP and POE injector works fine on my controller here at the shop.

I can remove any of the other 5 AP's that are on site and re-adopt them without issue.

I have "hard reset" the unit via the pin-hole.

I physically took the "bad AP" and the "new AP" to the networking equipment and hooked it up with 6ft patch cables at the switch - on yet another, different POE injector.... the same behavior. So, very doubtful it's a power or wiring issue.

Some of the Googling suggests that the unit isn't getting DNS or STUN server access, but the rest of the network seems to be in order (evidenced by the others not rebooting and the fact the unit does in fact work for a short period of time).

I'm leaning towards something wrong with the Unifi controller but I wanted to see if any of our resident Unifi guys have seen this before, and if you have, any ideas to fix it? Thanks!
 
Is it a L2 or L3 adoption you are doing?

Without adopting it ssh into the AP and try running a ping. It sounds like it may be failing to resolve the unifi name after adoption.

Sent from my SM-G870W using Tapatalk
 
You tried a different port on the switch? I see where you swapped out everything else, but I don't see where you mentioned that.
 
If you've got two WAPs in the same position with the same problems, and no other problems with other WAPs only the common points shared by the two WAPs are left to troubleshoot.

That is, switch port, patch cables, and wire in the wall. To me it sounds like you've got a damaged bit of cat 5 in the wall.
 
You tried a different port on the switch? I see where you swapped out everything else, but I don't see where you mentioned that.

That is, switch port, patch cables, and wire in the wall. To me it sounds like you've got a damaged bit of cat 5 in the wall.

"I physically took the 'bad AP' and the 'new AP' to the networking equipment and hooked it up with 6ft patch cables at the switch - on yet another, different POE injector.... the same behavior. So, very doubtful it's a power or wiring issue."

Yes, I used a different port and I eliminated the wiring as an issue by using patch cables at the switch (not at it's original install location).
 
...on yet another, different POE injector....

Yes, I used a different port and I eliminated the wiring as an issue by using patch cables at the switch (not at it's original install location).
Are you maybe saying POE injector, when you mean port-supplied POE? If that's the case, this would make sense, and it would explain why I missed that you tried a different port.
 
I'm wondering if the APs are getting an incorrect 'inform' URL from the controller. I've had that happen a few times. Have you tried adopting the APs using SSH?

I would try SSH-ing into one of the working APs first and type the command 'info' to get the inform url they're using.

Then reset one of the non-working APs and login using the default user/pass (ubnt/ubnt). Or login first and reset the AP over SSH:
Code:
syswrapper.sh restore-default

Once reset and logged in, enter the following command:
(The prompt changes to: UniFi#)

Then set the inform URL using the url you got from the working AP, eg:
Code:
set-inform http://10.10.10.10:8080/inform

Now try adopting the AP in the controller software.

I find that adoption is temperamental and can take a few attempts. If adoption keeps failing ('Disconnected'), something I've found that helps, is to try sending the set-inform command again while adoption is still pending, specifically around 25 seconds after starting adoption.
 
All of the above can be avoided if you have DNS control, because the default inform URI is unifi.whatever.local, based on the DNS suffix passed by the local DHCP server.
 
Well. Not sure what happened... after a week of not working/cycling on and off the damn thing started working sometime in the early morning hours today. It's been up for 12 hours.

I changed nothing. Bizarre.
 
This is why I prefer L3 adoption I just set it up and ssh in and do a set inform no messing with anything.

I've seen APs get temperamental if they are running firmware not supported by the controller version.

Sent from my SM-G870W using Tapatalk
 
Back
Top