I just switched out my UDM for a pfSense box, and am running the controller in a VM.
I opened ports TCP 8080, UDP 10001, 3478, & 1900 as floating inbound rules (I tried bidirectional, still didn’t help)
The issues are:
a) I am having is my controller sees all of my devices, but it is saying they are all disconnected.
b) It does not see any new or reset devices to adopt (I bought 1 new WAP)
Troubleshooting:
Reset one of be Beacons to see if I could adopt it. Unfortunately, now I can’t even see it.
Power cycled both switches (which also cycled the WAPs). No dice.
Put controller VM on same VLAN as UDM was. Still nothing.
If anyone has any advice and/or suggestions, please. I’m at a loss.
I’m no pfsense expert but i hope a flowing rule docent open those things to the internet
I recently did this exact same thing where i removed the udm in favour of pfsense.
When i did it i had to hold in the reset button on the back (which i assume you mean you did when you say you reset one of them) and it almost instantly showed up.
i would recommend having the controller and access points on the same VLAN just to make it easier for yourself and to make it have a better chance of working.
If you did both of those things I’m not sure if there’s anymore i can give you other than double check everything with the VLAN. Make sure the switch and DHCP are set up correctly and the routing rule is in place.
Unless you are tying to manage these devices remotely then they should be on the same network as the controller. If you are opening up to remote management the only two ports needed are 8080 and 3478.
The good news is, as per your advice, putting the controller on the same VLAN as the devices worked in terms of being able to attempt to adopt.
The bad news is that adoption keeps failing. I SSH’d into my switch and “set-inform” to point to new address of controller, but that is still not working
As for the floating rules, I deleted those since they are now on the same VLAN and redundant.
I just tried that with my main switch (via h/w button). Strangely, my new, unconfigured WAP is also failing to adopt as well.
It seems the default inform address is “http://unifi:8080/inform”, so I created a DNS record for ‘unfi’ to point to my controller. Worked as well as all of my other bright ideas. lol
I haven’t bothered resetting any other devices yet, as everything is connected to my 16-port switch, so I’ve been trying to get that one working first.
edit: I also created DHCP reservations for all of my devices in pfSense.
For clarification, these are the troubleshooting steps I have taken so far:
Reset my switches
Forgot all of my devices from my controller
Manually changed the ‘set-inform’ on my two switches & my new WAP
Added DNS records in my Pi-Hole for the default /inform address to point to my controller
Made sure controller & devices are on same network
Tried making port forwarding rules for any traffic on ports TCP 8080 & UDP 3478 to send to controller (deleted when didn’t solve issue)
Another odd behavior I just noticed, is my in-wall WAPs are still trying to adopt. Despite me resetting one of them, and forgetting them in the controller.
Unfortunately I can’t SSH into my controller, as it is outright refusing connections, even though I can access the web UI with no issues.
According to this subreddit post (if I’m understanding correctly), the issue is with mongodb not liking me restoring my UDM backup to a different device. In this case a VM hosting the controller, and that I have to manually purge each device that’s having problems (in my case, every one).
The link below explains how to do it in Windows.
Ideally, I would have liked to test it out and confirm before posting, but I hope this will prevent someone else from completely borking their network setup.
I think the only option left for me is to reinstall mongo & the UniFi controller & re-setup my network from scratch.
Thank you both @LTS_Tom and @originalusername for your help and advice, but I think it was too far gone. So I took the nuclear option and hooked up my UDM, so the VM could have internet access in order to reinstall the UniFi Controller. And so the rest of the family has internet access and not breathing down my neck. lol
Now I know what not to do. Restoring from a backup from different hardware (or at least from a UDM) is a hard no. I’ll need to set it up from scratch.
Update: I tried restoring one of my backups to my UDM (the source of the backups), and ran into the same issue, just not as bad. It appears to be the backup itself that is causing the issue. It seems as though the Mongo database is not correctly overwritten, thereby causing the adoption loop I was experiencing.
Anyone who has a lab environment, might want to test your backups to see if they will restore properly.