We had an UniFi controller installation where we were unable to adopt a small amount of devices, like three out of forty. It was very strange, everything worked perfecly except those few.
The error we found in /var/log/unifi/server.log was just:
server.log:[2019-06-20T16:11:53,707] INFO dev - [state] dev[aa:aa:aa:aa:aa:aa] UNKNOWN->INFORM_ERROR, state_expire=0
We tried manually updating the firmware of the switch via ssh, but there was just a minor version upgrade and it changed nothing.
Then I found an old thread on UniFi forums saying they solved a similar error by setting a manual IP for the hostname unifi in /etc/hosts. This looked a little weird to me because we were trying to inform using an IP address, but we tried everything so was worth a try.
So I added
10.11.11.1 unifi
in /etc/hosts, ran
set-inform http://10.11.11.1:8080/inform
At first on the controller I was still seeing Inform error, but after the second try I finally had Adopt required! Gotcha!
Can’t really tell why, but the /etc/hosts trick did it!
This was a huge help after pulling our hair out for an hour or so!
Thank you!
This particular issue didn’t help me, but what we ended up doing was using mongo to modify the MAC address in the Unifi database so that it didn’t “know” about our switch anymore. Then we factory reset the switch, adopted it as new, then copied the config from the “old” switch with the bogus MAC.
Maybe this will help somebody else – these instructions are for a Linux controller, but should still apply nicely to another platform:
Run the mongo client:
mongo –port 27117
Switch to the ace database:
use ace;
Update the database to a different MAC (I just changed the last byte to ff):
db.device.updateOne({“mac”:”80:2a:a8:9b:62:c1″}, {$set: {“mac” : “80:2a:a8:9b:62:ff”}})
After the update is acknowledged, press CTRL-D to end your mongo session, then restart the unifi/mongo services:
service unifi restart
Then you go into the controller, adopt the “new” switch, copy the config from the “old switch”, and then delete the “old switch”. You MAY have to ssh in and factory reset the switch first, but that is easily Google-able.