03-17-2018 05:15 PM
It's not really a big issue and not bothering me as such, except what is generating these events? It's another weird Ubiquiti/Unifi quirk that makes me question why I spent so much on this kit, it all seems so unfinished and unprofessional - my cloud key appears to have randomly rebooted a few hours ago too with no reason, always something odd going on. Here is just one example, logs are being filled up showing clients disconnect and reconnect sometimes for minutes at a time apparently but as far as I can tell, they aren't.
They all appear to be clients connected via the 24 port non PoE switch, I have 2 other 8 port switches and no disconnects listed. I've ran continous pings to the Dell ESXi host shown in this example and no packet loss at all, no alerts or alarms on the host, so where are they coming from and why? They are all showing 0 Bytes too, though sometimes it does show files in the KB range.
Switch running 22.214.171.12491, Unifi is 5.7.20 and switch was restarted yesterday when the latest versions were installed and it was showing these events before the update and before the restart, infact they go back weeks.
03-17-2018 06:52 PM
I have a Dell PowerEdge T20 ESXi host that was doing the same.
I posted about these "disconnect" events earlier... Wired Client Random Disconnect Event Logs - it's quite irritating as my Unifi CCTV cameras are doing the same now.
03-18-2018 12:52 AM
Interesting, are yours actually disconnecting or just reporting that they are? As far as I can tell everything is still connected and working fine, as I said continous ping results in zero packet loss.
03-18-2018 02:33 AM
No, they are not actually disconnecting at all. The cameras continue to record during the times between being reported as disconnecting and reconnecting. The Unifi NVR correctly lists the uptime of the cameras. It’s only the Unifi network controller that reports the supposed dis/reconnects.
03-18-2018 02:38 AM
Oh, one other thing I meant to say. If you have been running persistent pings to the devices, that would rule out the MAC address timeouts that UBNT suggested as the reason in the posts I linked to in mine.
04-17-2018 03:11 AM
Also experiencing this same behavior.
I noticed that my ESXI host does Not show up in the clients listing until I login to the the host through the ESXi Managment Console, after which is shows as a listed client with the correct IP address.
My Synology DS1815+ NAS also behaves in a similar fashion. The common thread being they both have dual uplinks to the US-48W-POE switch. The ESXi is running an Active/Active using the same static IP while the Synology NAS has 2 active uplinks with different static IPs assigned. The NAS will show up in the clients list but it only lists 1 of the 2 active uplinks until an interactive login occurs on the device itself????
04-22-2018 07:21 AM
Did anyone find a resolution to this? The events are pretty annoying, I looked back through my events and before the latest firmware upgrade on the switch it wasn't having the issue. So I downgraded the firmware and no success still same annoying error. Tried different cable as well as I thought it could be that. Only thing I didn't do was downgrade cloud key from the 5.7.
08-25-2018 05:45 PM
I'm seeing exactly the same behaviour for two of my Hyper-V server NICs (dedicated Hyper-V NICs, not shared with host).
It's not a real issue, but very annoying and I was wondering if someone found a solution for this?
08-25-2018 06:32 PM - edited 08-25-2018 06:34 PM
I don't know if this was ever solved on this thread. I had this issue recently and found this:
If I SSH into my hardware and type the info command, I find that the set-inform status URL is not defined on the hardware to my remote controller. It is actually shows as http://unifi:8080/inform on info response.
Once I set the inform on every piece of hardware via SSH using the controller's WAN IP address and not the FQDN, I stopped getting the connect/disconnect notices. Also if I login and SSH again to my hardware and info, I now get the set-inform status as connected to my controller.
My original addoption for these devices were done by using the iPhone App of the Chrome App on my PC.
NOTE: My controller is on a remote server, so it is not via a cloudkey, or in-house server.
08-26-2018 06:12 AM
Good tip...I reviewed and confirmed that my UNIFI hardware already has the correct set-inform info set. That said what I was previously seeing as disconnect/connect has corrected itself for all clients "Except" my physical ESXi host???
Not really a problem to this point, more of just an annoyance that I was intrigued by.
09-03-2018 02:36 PM - edited 09-03-2018 02:38 PM
It's not actually my Unifi wifi kit/switches that show the disconnects - it's my Unifi CCTV cameras and my ESXi host - the Unifi cameras have no need of the inform settings.
None of them are actually disconnecting and re-connecting - the cameras are always up and the VMs running under ESXi are always available.
The cameras used to be patched into my USW 8-150 and did the same - they are now on an Edge switch 8-150 and show the smae symptoms.
in the thread I originally linked to, I was advised to try a purge of the Unifi controller on my cloud key and a settings only restore which I couldn't be bothered to try at the time. I have recently done that after an attempted upgrade to the 5.9.x beta controller and sunsequently needed to go back to the 5.8.x stable branch. As I expected, it made not a blibd bit of difference to the reported disconnections.
They are pretty irritating as I have 3 CCTV cameras plus the ESXi host reporting the events constantly.
09-30-2018 05:47 AM
Big thanks for the hints and sorry for my late reply!
In my case (Hyper-V, IPv4), the "issue" could be solved by sharing the affected NICs with the host and disabling IPv6 on each adapter...
a week ago
I've been getting the same errors. Plus, for some reason, two of my UniFi downlink switches are now showing up as clients but only showing their MAC address under the client page of my controller. I'm running a Gen 2 Plus CK, and this just started happening after updating to the v5.10.17 controller a couple days ago.