Unifi Problems

[tri_94]

Hi there we have had a Unifi Wireless System for the last 2 or more years. Everything was running perfectly until a few weeks ago when suddenly it has become really unstable and flaky. Pings can be anything from 1ms - 3000ms and have lots of drops outs.

 

We hadn't touched anything on the Unifi controller. Since the issue we have replaced the software based controller with a cloud key and updated the latest versions and now even the newest beta versions of both controller and AP firmware.

I've spoken to Unifi who told me to set all AP's on Low power and change the channels manually. Even change HT from Auto and also Encryption settings. But nothing seems to help. If a reboot an AP get a bit a few lower point for a short while. Also don't know if anyone else has seen issues like this but IOS Devices can seem to kill connection. If an Stock iPhone 6 or 6s is connected to our test AP the pings raise to the 000's and then request timed out... disconnect the phone and pings drop to 1-10ms for a short period of time.

 

With have 68 AP's and 100 end devices that should be using our system.

 

Any ideas would be great.

 

Strange thing is the only thing that I can think happened before the problems started is a powercut.

 

Thanks

[sted]

is it worth powering down all the aps then bringing them up one by one? maybe one is doing something daft

[Meldrew]

So there's no difference if a client connects to different APs - they all seem to do the same, no matter which you connect through? What sort of wireless network are you using? 802.1x, pre shared key? If you setup a continuous ping to a phone (for example) and then restart the phone, do the pings remain consistently bad immediately after bootup or does it start good then deteriorate?

 

Meldrew

[tri_94]

is it worth powering down all the aps then bringing them up one by one? maybe one is doing something daft

 

Hi there tried that and we reboot all AP's everyday.

[tri_94]

All the AP's seem the same and have even tried a different controller with a single ap and the problem still seems to be there. Tried open to see its if was the WPA2 causing an issue. Currently have two laptops connected on a pro AP on 5g which is the only 5g Signal and pinging the gateway on one. Nothing else using the AP and pings range from request timed out down to 4ms

 

 

 

So there's no difference if a client connects to different APs - they all seem to do the same, no matter which you connect through? What sort of wireless network are you using? 802.1x, pre shared key? If you setup a continuous ping to a phone (for example) and then restart the phone, do the pings remain consistently bad immediately after bootup or does it start good then deteriorate?

 

Meldrew

[dapaulio]

 

Strange thing is the only thing that I can think happened before the problems started is a powercut.

 

Thanks

 

sounds like you have a device chatting gibberish and Is swamping your network. Check your switch logs for any abnormalities. this may help you narrow down the problematic device (or devices) try switching off all your POE devices. restart the switches and power the devices up one at a time.

[tri_94]

sounds like you have a device chatting gibberish and Is swamping your network. Check your switch logs for any abnormalities. this may help you narrow down the problematic device (or devices) try switching off all your POE devices. restart the switches and power the devices up one at a time.

 

Any easy way to discover if this is the case?

[sted]

Any easy way to discover if this is the case?

depends how nasty to users you want to be. You could disable the current ssid/password create a new one and slowly reconnect devices (or do it when most people have gone home and just use a few test devices see if the problem persists with very few devices connected)

[tri_94]

depends how nasty to users you want to be. You could disable the current ssid/password create a new one and slowly reconnect devices (or do it when most people have gone home and just use a few test devices see if the problem persists with very few devices connected)

 

Well I've tried with an AP that has a new SSID with WPA2 and no one knows the key.

 

Interestingly have and old net gear AP which I tested and pings were fine from that normally 1-30ms a few higher ones but nothing like our site wide Unifi system

[JATSO]

I would block all non school equipment and inform the staff you have an issue and see what happens after a reboot. Only let staff mobiles etc onto the guest network.

[tri_94]

I would block all non school equipment and inform the staff you have an issue and see what happens after a reboot. Only let staff mobiles etc onto the guest network.

 

I've blocked about 300 devices already including some staff. Strange thing was never needed to do this before and it was working fine. And again new SSID on a software controller WPA2 with no one else with the key and still having the same issues.

[dapaulio]

Any easy way to discover if this is the case?

 

Sounds like a double patch or an affect caused by the powercut.

 

I meant the devices physically attached to your switches things like WAPs and phones.

 

as you mentioned you had a powercut it is more likely to be that rather than a double patch which may have done something to an end device and it is chatting gibberish. too narrow down the devices I would log on to each switch check the port logs for errors which will help you identify possible faulty devices, then restart the switches. if it is a faulty device you should find your network will stabilize momentarily then begin to show pings again in the 000's. which ever one of the switches you last rebooted will have the faulty device on it. next unplug all the devices on that switch, reboot the switch again and your network should stabilize again. may be wait 10 mins to be sure it remains stable then begin plugging the devices connected to that switch in one by one a minute between each. as soon as you network starts to show elevated pings that you can narrow it down to within a few device on the other end of those cables.

[sted]

Well I've tried with an AP that has a new SSID with WPA2 and no one knows the key.

 

Interestingly have and old net gear AP which I tested and pings were fine from that normally 1-30ms a few higher ones but nothing like our site wide Unifi system

are you still on the same network? might be worth setting up a self contained network with no links to your main one

[JATSO]

Could it be just a rogue AP or your switch that the APs are connected via.

[box_l]

Are you assigning specific channels to the APs?

 

If you boot them all at the same time the can all come up on channel 1 for instance.

 

Try rebooting them one after another with a 5 minute gap between them and see if that solves it.

[tri_94]

Thank for that but I can see on the controller that they are on different channels. We have some over lapping but when I can plug and old netgear unit and get a far better connection and it will give me far lower pings.

 

Are you assigning specific channels to the APs?

 

If you boot them all at the same time the can all come up on channel 1 for instance.

 

Try rebooting them one after another with a 5 minute gap between them and see if that solves it.

[tri_94]

Could it be just a rogue AP or your switch that the APs are connected via.

 

No sure about rouge AP as I've tested with a single AP on a different controller and still have that issue. As for the switch I can plug my old net gear ap in the same switch as my Unifi unit and get far lower pings and a better connection.

[tri_94]

Would enabling QOS help the problem do you think? If so anyone have any idea about QOS on HP1910's ?

[tri_94]

Thanks to everyone that's replied so far.

 

After reading all your posts I've setup a single switch with an access point and controller and connected two laptops one via wireless and other cabled. Testing pings from one to the other and pings are fine. I'm getting 1-15ms with a rare higher one. When I plug in my network to the switch I get 70ms upwards to 600ms a few request timed outs... I then unplugged the network and pings drops and after a couple of minutes back down to 1-15ms with a few more higher pings then before. Its as if its stabilizing as longer I leave it it seems the less frequency the higher pings are.

 

So it appears to be network related and not AP's 90% of my switches are HP1910 with STP enabled which I know for a fact if I plug a loop cable in it shuts down the port. Have tested this to the point I can have over 18 loops in one switch and the switch operates fine. Problem is we have little if any downtime. Any ideas how I can address this without downtime? As i'd like to split the network and power up switch / area at a time and see when the problem occurs but this isn't really possible

 

Thanks

 

 

 

 

 

Any ideas

[psydii]

Wireshark.

 

Put a computer into promiscuous mode and see what is flying about wired-side. it only takes a few MB/sec of broadcast traffic to bring a 802.11n network, while you wired-side would barely notice.

 

If you have a macbook pro you can do the same on the wireless side.

 

If you are unfamiliar with wireshark all you really are doing is look for patterns and anomalies. (tons of BROADCAST MULTICAST ARP REQUESTS or STP packets are the easiest to spot). So perhaps start by watching your *isolated* switch and the attached AP, then watch what changes when you patch the isolated switch into the network.

[tri_94]

I've looked at Wireshark before but not sure how to filter out that data I don't need. I have used Colasoft Enterprise trial and noticed some Broadcast ARP Requests mainly from our Print Server but had read that could be normal?

 

 

 

Wireshark.

 

Put a computer into promiscuous mode and see what is flying about wired-side. it only takes a few MB/sec of broadcast traffic to bring a 802.11n network, while you wired-side would barely notice.

 

If you have a macbook pro you can do the same on the wireless side.

 

If you are unfamiliar with wireshark all you really are doing is look for patterns and anomalies. (tons of BROADCAST MULTICAST ARP REQUESTS or STP packets are the easiest to spot). So perhaps start by watching your *isolated* switch and the attached AP, then watch what changes when you patch the isolated switch into the network.

[sted]

I've looked at Wireshark before but not sure how to filter out that data I don't need. I have used Colasoft Enterprise trial and noticed some Broadcast ARP Requests mainly from our Print Server but had read that could be normal?

unplug it for 5-10 mins see if the wireless improves one less thing to worry about

[LeMarchand]

unplug it for 5-10 mins see if the wireless improves one less thing to worry about

 

But I need to print (my Pizza Express voucher) NOW!!!

[psydii]

I've looked at Wireshark before but not sure how to filter out that data I don't need. I have used Colasoft Enterprise trial and noticed some Broadcast ARP Requests mainly from our Print Server but had read that could be normal?

 

"some" Yes likley normal. Hundred per second, no not normal. You print server is likely polling the printers.

 

This point of this particular exercise is to see **ALL** the traffic. If there is a flood of some kind it will drown out all the other background 'noise'. If you are only seeing "who has ip xxx.xxx.xxx.xxx tell nnn.nnn.nnn.nnn" packets, then you are not in promiscious mode. Mind you, if there are only a few per seocnd then this shouldn't be a problem.... UNLESS for some reason all clients wireless-side are responding, inwhich case that could be flooding the wifi, and leaving the wire-side relatively un-impacted.

 

Perhaps a better refinement would be to mirror a port your AP is connected to and watch that. If you see 50-100Mb/s traffic in either direction your wifi is being flooded.

Edited July 8, 2016 by psydii

[JATSO]

I had an issue recently where the patch leads that connect all the switches were causing issues so I changed them for some better quality made CAT6 patch leads.

 

Sorted out my connection and drop out problems.

Edited July 8, 2016 by JATSO