HCHTech
Well-Known Member
- Reaction score
- 3,823
- Location
- Pittsburgh, PA - USA
I'm working with a new residential customer doing a home remodel who had an existing unifi system consisting of 3 AC-LR APs and one of the outdoor units with the rabbit ears. 2 of the APs are wired with home runs, and the other ones are meshed. We met back in August, where I reviewed what equipment was needed and the client bought it directly from the Unifi store. The existing wired APs were powered with PoE injectors, so as part of the re-do, I suggested a small PoE switch, the USW Lite-8-PoE, since it has 4 PoE+ ports and the 52W budget would work for the suggested final setup.
We're utilizing all existing wiring, including some runs that weren't previously terminated at all, just coiled in boxes at both ends. We're adding 3 new APs, a U6 Pro, an InWall 6, and a U6-Mesh (the toilet paper tube - see, I remembered @YeOldeStonecat!)
Because of the limitations of the wiring, we still have only 3 devices powered by the PoE switch, but one of them is the InWall6 - which has a single PoE+ port OUT. So we have the InWall6 powered by a single PoE line, and want to use the output to power another AP using existing wiring that goes from the InWall location to a new addition.
Yesterday was "put it all together" day, and of course, some problems were discovered.
There is no PoE output on the InWall6 unit, even though it is lit up just fine and reporting to the controller as it should (Hostifi). We solved this by plugging in a PoE Injector on that line out from the InWall6, although I read the specs of the PoE switch and InWall several times and I think it should work.
The switch adopted just fine to the controller, but a few minutes later, even though it was working, it started showing as offline in the controller.
I went ahead and adopted the rest of the new APs and updated their firmwares, then factory reset the old APs and started to adopt them. The Outdoor unit adopted and updated fine, but the AC-LRs were so much trouble - I factory reset them multiple times and power cycled them, but they would never show up on the network, so I couldn't SSH into them to do the adoption. Finally, as a last act before giving up for the day, I power cycled the switch, and one of those LRs immediately popped up ready for adoption - weird. I successfully adopted and updated it, but one more attempt on the final LR was unsuccessful, so I gave up and took it back to the shop to see if I could adopt it there.
This morning on my bench, I setup a little router for the job, and did a factory reset on that LR, it showed up on the network right away and I successfully adopted it to the client's site in the controller. Hmm. This makes me think something is going on with that little PoE switch.
I went back to the clients today, plugged in the final LR and it immediately meshed with the network (it is one that doesn't have a home run in its location).
For the Switch, Everything was working, but I couldn't detect it with an IP scan or ping it at its last known address, so I deleted it from the controller and factory reset it...which did not go well. It would not show up on the network, so I couldn't determine its IP address. Multiple scans with an IP Scanner did not show a device with a matching MAC address. So I think there for sure is something wrong with it. Finally, after power-cycling it a couple of times, it went into TFTP mode - ok, so possibly corrupted firmware. While I was downloading the TFTP client to see if I could overwrite the firmware, the client turned off a breaker as part of some other thing he was working on, and the power was cut. When it came back on, the switch suddenly had an IP address and it was clear it had been factory reset (as I was able to SSH into it with the ubnt/ubnt credentials. I adopted it successfully an then everything seemed to be working. I wasn't confident and told the client as much, but added up my time, got paid and out of there for today at least. I promised to do the tuning of the APs over the weekend and then we will talk again on Monday about next steps.
By the time I got home 30 minutes later, I pulled up the client on my controller and sure enough, the damned switch was reporting offline again. All of the APs were online and I didn't see anything else concerning. I'm pretty confident now that the switch is bad.
I've already ordered a replacement switch - I needed one for stock anyway. When it comes in, I'm going to swap it out and see if that solves the problems. If so, the client can pay me for the new switch and RMA the old one with Unifi on his nickel.
For unifi folks - have you ever run into this behavior? I guess I've been lucky, I've got about 50 clients on my controller now, and a few hundred APs, switches & such, and I've never had trouble like this.
Also - Am I correct or not that the OUTPUT PoE port of the InWall6 should power a U6Pro? The documentation sure seems clear enough that it's PoE+ output as long as it is PoE+ input. This could all come back to a problem with that little PoE switch, too. What a headache. $100 part, $200 in labor to screw around long enough to decide it's bad. At least the client bought the thing, which isn't my normal modus operandi, but that makes him on the hook for the RMA - small favors.
We're utilizing all existing wiring, including some runs that weren't previously terminated at all, just coiled in boxes at both ends. We're adding 3 new APs, a U6 Pro, an InWall 6, and a U6-Mesh (the toilet paper tube - see, I remembered @YeOldeStonecat!)
Because of the limitations of the wiring, we still have only 3 devices powered by the PoE switch, but one of them is the InWall6 - which has a single PoE+ port OUT. So we have the InWall6 powered by a single PoE line, and want to use the output to power another AP using existing wiring that goes from the InWall location to a new addition.
Yesterday was "put it all together" day, and of course, some problems were discovered.
There is no PoE output on the InWall6 unit, even though it is lit up just fine and reporting to the controller as it should (Hostifi). We solved this by plugging in a PoE Injector on that line out from the InWall6, although I read the specs of the PoE switch and InWall several times and I think it should work.
The switch adopted just fine to the controller, but a few minutes later, even though it was working, it started showing as offline in the controller.
I went ahead and adopted the rest of the new APs and updated their firmwares, then factory reset the old APs and started to adopt them. The Outdoor unit adopted and updated fine, but the AC-LRs were so much trouble - I factory reset them multiple times and power cycled them, but they would never show up on the network, so I couldn't SSH into them to do the adoption. Finally, as a last act before giving up for the day, I power cycled the switch, and one of those LRs immediately popped up ready for adoption - weird. I successfully adopted and updated it, but one more attempt on the final LR was unsuccessful, so I gave up and took it back to the shop to see if I could adopt it there.
This morning on my bench, I setup a little router for the job, and did a factory reset on that LR, it showed up on the network right away and I successfully adopted it to the client's site in the controller. Hmm. This makes me think something is going on with that little PoE switch.
I went back to the clients today, plugged in the final LR and it immediately meshed with the network (it is one that doesn't have a home run in its location).
For the Switch, Everything was working, but I couldn't detect it with an IP scan or ping it at its last known address, so I deleted it from the controller and factory reset it...which did not go well. It would not show up on the network, so I couldn't determine its IP address. Multiple scans with an IP Scanner did not show a device with a matching MAC address. So I think there for sure is something wrong with it. Finally, after power-cycling it a couple of times, it went into TFTP mode - ok, so possibly corrupted firmware. While I was downloading the TFTP client to see if I could overwrite the firmware, the client turned off a breaker as part of some other thing he was working on, and the power was cut. When it came back on, the switch suddenly had an IP address and it was clear it had been factory reset (as I was able to SSH into it with the ubnt/ubnt credentials. I adopted it successfully an then everything seemed to be working. I wasn't confident and told the client as much, but added up my time, got paid and out of there for today at least. I promised to do the tuning of the APs over the weekend and then we will talk again on Monday about next steps.
By the time I got home 30 minutes later, I pulled up the client on my controller and sure enough, the damned switch was reporting offline again. All of the APs were online and I didn't see anything else concerning. I'm pretty confident now that the switch is bad.
I've already ordered a replacement switch - I needed one for stock anyway. When it comes in, I'm going to swap it out and see if that solves the problems. If so, the client can pay me for the new switch and RMA the old one with Unifi on his nickel.
For unifi folks - have you ever run into this behavior? I guess I've been lucky, I've got about 50 clients on my controller now, and a few hundred APs, switches & such, and I've never had trouble like this.
Also - Am I correct or not that the OUTPUT PoE port of the InWall6 should power a U6Pro? The documentation sure seems clear enough that it's PoE+ output as long as it is PoE+ input. This could all come back to a problem with that little PoE switch, too. What a headache. $100 part, $200 in labor to screw around long enough to decide it's bad. At least the client bought the thing, which isn't my normal modus operandi, but that makes him on the hook for the RMA - small favors.
