AC87U: 100% CPU load on both cores and connection problems

[yikyi]

Hi there

I am running a AC87U (Merlin 384.13) with a couple of additional scripts running:

• DNScrypt
• Skynet
• diversion
• (pixelserv)
• (amtm)
• (swap enabled on USB2)
• (RNG)

Very often I have Wi-Fi issues. My smartphone or my laptop say that they are connected to the network, but have no internet connectivity. However, in these cases I canoot access the WebUI of the router. Today the XBOX1 said it couldn't connect to the DHCP-server. Most often only few devices are affected. Some work, at the same time as others don't.
Just now I saw that the CPU load is very high, above 80% on both cores for most of the time. Is there a way to check what causes the high load? Or is any of the above extensions already known to be quite consuming?

Thank you very much in advance for your hints and guesses!

 

[grifo]

To see what is causing the high cpu you have to ssh to the router, type "top" and hit enter. Some cpu spikes are normal but if they last for minutes there can be a problem that could well be caused by one of the scripts.

Otherwise the type of random problems you're experiencing will often get fixed with a factory reset and a manual reconfiguration of the router, I know it's a PITA but more often than not that'll do it. Has it been long since you did one?

I also have an AC87U running 384.13_1, Skynet, Diversion, Pixelserv, amtm and other scripts and it doesn't have any of those problems.

 

[WNorkle]

Unlike grifo, my AC87u has regular occurrences of both CPUs at 100% for some period of time which will eventually result in a loss of DNS. Restarting dnsmasq sometimes helps, mostly doesn't. Regular restarts seem to make the most positive difference.

To troubleshoot, I have been removing scripts - and now just have diversion/skynet/YazFi/amtm/scMerlin. Just playing the waiting game.

 

[yikyi]

To see what is causing the high cpu you have to ssh to the router, type "top" and hit enter. Some cpu spikes are normal but if they last for minutes there can be a problem that could well be caused by one of the scripts.

Is there also a possibility to create such a log-file? Because, if I cannot connect to the network, I am not able to establish a SSH connection.

Unlike grifo, my AC87u has regular occurrences of both CPUs at 100% for some period of time which will eventually result in a loss of DNS. Restarting dnsmasq sometimes helps, mostly doesn't. Regular restarts seem to make the most positive difference.

I set up a daily reboot during the night now. Maybe this helps. Only time will tell. Thank you for the idea! I'll try this over the week now and see if the situation improves.

 

[weslsew]

are you using ai protection? if so, try disabling it

 

[grifo]

Is there also a possibility to create such a log-file? Because, if I cannot connect to the network, I am not able to establish a SSH connection.

If you couldn't connect to the router while the cpu was spiking above 80% how could you see it was doing that? You wrote that some devices are affected while at the same time others aren't, then ssh to the router from one that isn't.

I set up a daily reboot during the night now. Maybe this helps. Only time will tell. Thank you for the idea! I'll try this over the week now and see if the situation improves.

If I were you I would have already done a factory reset rather than continuing to suffer with these problems for days. If you decide to do the reset search the forum for the best practices to do it, for example how to save your static DHCP assignments.

 

[yikyi]

Now, I happened to be connected to the WiFi while the CPU load in the webUI was around 100% again. But "top" lists not really a service that causes such a high load. :S

Mem: 225324K used, 30352K free, 32K shrd, 1116K buff, 5448K cached
CPU: 1.6% usr 5.2% sys 0.0% nic 91.7% idle 0.8% io 0.0% irq 0.3% sirq
Load average: 2.21 1.92 1.33 1/138 28805
PID PPID USER STAT VSZ %VSZ CPU %CPU COMMAND
27 2 x SW 0 0.0 0 1.5 [mtdblock3]
290 261 x S 5060 1.9 1 1.3 nt_monitor
263 261 x S 5060 1.9 1 0.9 nt_monitor
265 1 x R 5192 2.0 0 0.7 nt_center
326 1 x S 6552 2.5 0 0.3 httpd -i br0
252 1 x S 2604 1.0 1 0.2 protect_srv
351 1 x S 6552 2.5 1 0.1 networkmap --bootwait
354 1 x S 5560 2.1 0 0.1 mastiff
331 1 x S 5396 2.1 1 0.1 watchdog02
142 1 nobody S 784m313.7 1 0.0 /jffs/dnscrypt/dnscrypt-proxy -syslog -config /jffs/dnscrypt/dnscrypt-proxy.toml -child
1308 1 nobody S 109m 43.8 1 0.0 dnsmasq --log-async
1368 1 x S 7524 2.9 0 0.0 minidlna -f /etc/minidlna.conf -r
1371 1370 x S N 7524 2.9 1 0.0 minidlna -f /etc/minidlna.conf -r
1370 1368 x S 7524 2.9 1 0.0 minidlna -f /etc/minidlna.conf -r​

nt-monitor and dnsmasw --log-async are the two services that are most of the time above 1% CPU load, sometimes above 2%, but no where near 100%. There are many more things listed with 0.0% CPU load. Are there addition things that aren't listed at all, which could cause a high load?

Thank you for your replies.

 

[grifo]

It looks like it was a spike and by the time you issued the command it had already gone, top is showing the cpu 91.7% idle. Your memory usage is high though.

Try removing DNScrypt and use the firmware built-in DNS-over-TLS.