[RT-AC3200] Router unstable

[ChrisL8]

I purchased the RT-AC3200 router in the hopes that it would provide a stable platform for my home network.
It was probably overkill, but my hope was that by starting with the best I could relax and work without fussing with my router every day.

As soon as I got the router I added the RT-AC3200_3.0.0.4_378_7838 firmware and factory reset it.

The first day I set it up it kept "disappearing". Everything would be working fine and then it would just drop out.
The 5Ghz and 2.4Ghz signals were there, but nothing could authenticate with them.
I could not PING the IP of the router or see any other ports on it.
The lights on the router kept blinking like it was doing its job, but it was unresponsive.
The only solution was to power cycle it.
Upon booting it the logs gave no indication of anything unusual.

This happened like 5 times on the first day,
then less the next.
I kept shutting off features like the AiCloud stuff, the USB Applications and the AiProtection stuff.
Eventually it seemed to get stable.

However, it still does this at least once a week.

I upgraded to the 3.0.0.4.378_9135 firmware when it came out, just because it didn't seem like it could hurt, but it has not improved the disappearing acts.

For a >$200 router this is very frustrating to have it go silent like this.

Does anyone have any suggestions on how to diagnose, fix or improve this?
Should I return it or attempt a warranty replacement?

Thank you for any suggestions, I'll certainly post feedback on anything I find to help.

P.S. I also found the Smart Connect to be useless, so I disabled that. A bit of a disappointment, but better than having clients disconnected periodically.

 

[thiggins]

Shouldn't do that. I'd return it and try another one.

AC3200 class only helps if you have a lot of 5 GHz clients. Otherwise it's the same as an AC1900 class router.

 

[ChrisL8]

8 seems like a lot of 5 GHz clients to me, but I don't know. With my previous 802.11N 5 GHz setup if I did any streaming, remote desktop sharing or large data transfers it caused other people on WiFi to have trouble with their connections.

The Smart Connect is useless though. The phones love it, Windows machines seem to have mixed results, but it would disconnect VPN sessions randomly and my Linux systems would get kicked off of the network completely at random times. So I just setup 3 SSIDs and distribute the systems by hand. Basically just put the phones on one 5 GHz network and the laptops on another. I'm not sure if that makes sense, but at least the two won't bother each other.
The downside is that no one (except the ones without 5 GHz radios) gets to use 2.4 GHz this way, which does seem to have more range. Sticking with one network per device seems to be easier though.

 

[Puppa]

With a large set of clients, the cpu load on the 3200 will peg at 100% causing some odd failures. Starting from the factory defaults, what features have you turned on? On mine, I have turned off a number of items to get my large home network (> 35 clients) stable. You can watch the cpu load on the main page of the router interface under status. I have found that if the load is frequently above 40% on either of the two cpu's, there will be stability issues.

And no, there is no other home router brand or model yet that can handle my network unless I turn features off. Some of them cannot even handle it if I turn everything off, save for basic routing functions. Asus (and the 3200) comes the closest, however.

 

[ChrisL8]

I have one USB 3.0 drive attached, although it isn't used a lot.
No "guest" networks.
Everything under "AiProtection" (Network Protection and Parental Controls" is off.
Under Adaptive QoS "Apps analysis" is ON, but QoS itself is OFF.
Traffic Analyzer is ON.
Everything under AiCloud 2.0 is OFF.
Smart Connect is OFF.
DHCP is enabled.
IPv6 is OFF.
UPnP is OFF.

Basically it isn't doing much of anything.
I would like to turn on some features, but I've been waiting to feel good about the stability first.
It hasn't crapped out again since Saturday's power cycle.

Perhaps I should set up a cron job on another system to grab CPU stats from the router every few minutes to see if it trends upward at some point.

 

[Puppa]

Traffic Analyzer and Apps Analysis stand out on this list. As those are purely informational, they can be turned off. If you play video games, turn on UPnP. It's also safe to turn on QoS, but I always make sure to put realistic values in there. If you set your uplink throughput too low, the router will prevent you from attaining highest possible uplink throughput. On the other hand, if you set uplink throughput too high, the router will starve some clients and that will cause additional client requests.

It's not necessary to do a cron job--just get a few things streaming simultaneously and uploading/downloading large files and see how the load looks. Also, browse your attached disk to make sure that's not a major culprit (it could be). If either or both cpus peg upwards of 80% under stress, you're close to if not in the danger zone.

 

[ChrisL8]

Thank you very much for the tips. It helps to have something to go after.

I turned UPnP back on. I was aware of the benefits, but afraid it might be a security risk and a possible culprit in my issues.

Traffic Analyzer was nice for figuring out what clients were hogging the Internet, but information is less important than actual throughput and reliability.
I turned off Traffic Analyzer and the memory usage went from a steady 80% to 20%. That is interesting.
I also noticed that when I just clicked on the Traffic Analyzer page both CPUs spike to 100% for a few seconds!
I reset the router to clear the traffic logs and now it doesn't do that, and it shouldn't generate any new traffic logs since that is off.
Perhaps this Traffic Analyzer is a bad deal.

I also notice that the CPU spikes to 100% briefly when I click on the "Adaptive QoS" page, even though the "Apps analysis" is actually off.

Running a full backup of the USB drive over the network does push one CPU up to between 70 and 80% and hold it there. The other CPU stays below 15% though.
I hate to drop this feature, as we use it to pass files around the place, but I'll keep an eye on it and drop it if I have another crash.
Memory usage also goes up when I access the USB drive, and then drops again, not immediately, but after a while. I assume some caching is going on.

I'm going to wait until it is stable for a while longer before experimenting with QoS.