I have a N100 topton router running openwrt and 2 Ubiquiti APs and I get latency spikes up to 100ms after about 5 mins of a device being connected to one of the APs. Like when I run ping it goes from 3ms to 100ms back to 3ms and so on. Just wondering what would be the best way to diagnose this issue as it has only just recently started happening and I have tried reverting all the changes I've made of recent but it's still happening.
I will also test to see if this is happening with a device directly connected via Ethernet.
Here is info from the n100 router
The two unifi APs are a U7 pro max and U6 enterprise.
And I'm not using the router as an AP.
I'm fairly certain it's a misconfiguration of the router or another device on the network flooding it or something.
The latency spikes are measured on wifi connected devices?
If yes, are these battery powered? Then this might be coming from powersaving features.
If these are apple devices AWDL (I hope I got this right) is known to cause latency spikes...
Yeah I did initially think it might be power saving mode but I don't think so as desktops also have same issue
I also didn't touch my unifi config since it started happening only updated one of the APs but I have tried downgrading it and factory resetting and didn't fix the issue.
Also I don't think power saving mode would cause a latency spike every other ping
I also have uapsd off
I will see if I get latency spikes on a Ethernet connected device which should rule out any issue with WiFi and the APs.
I have also read on Reddit that someone else has had a similar issue with unifi and turns out it was because a nvr was flooding multicast or something.
I have tried disabling multicast for my network bridge and all ports and it didn't help however
You have desktop computers connected via WiFi? Anyway these powersavingbfeatures I am talking about tend to live in the WiFi chipsets for stations, so might well affed WiFi adapter for desktop systems as well...
That would be compatible with the idea, that it is the stations, not the APs that are responsible...
Ah, that is new information... could you post the output of say 30 seconds of ping over one of the affected paths please?
That would a good test, please report both success or failure...
Mmmh, I probably would have tried a packet capture first to see whether there are multicast packets to beginn with
Yeah the weird thing is with one AP ping to the router is fine but you can see the spikes in pings to Google for example. Where the other AP the spikes are in both pings to router and the Internet.
Can you mtr , if ping spikes occur between your internet gateway (N100?) and providers infrastructure even radical QoS will not save you from their oversubscription.
I have starlink and run cake autorate to get highest possible speed with lowest latency possible. It has been fine up until recently. Plus latency from router to internet is fine it's only from clients to router on one AP and client to internet on both APs.
I disabled auto neg on all my ports and unplugged the nvr I have on my network and I haven't had the latency issue again. Yet...
Also tried plugging my phone in directly to router via Ethernet and latency is high for Ethernet but not spiking but speed is garbage. Only getting 350mbps through openspeedtest despite the 1gbe dongle.
Here is ping ran on a desktop connected to same AP as last one. Weird goes up to about 100ms then slowly comes down. Thinking it could be APs again but honestly have no idea what settings to change to fix it I have changed quite literally everything in the unifi controller which is one reason I ruled out it being the APs.
Please can you try to export log file from cake-autorate as per GitHub README here? Then we can analyse the latency and see whether it correlates with load.