384.9 ROM Crash

[nihon]

Hi
Since I upgraded to 384.9 my Asus RT68U crashes every second day.

If I was connected to the Wifi I still am connected but with no internet access. If I try to disconnect and then connect again I get authentication error.

Restarting the router (power reboot) solves the issue for another two days.

I can see that the CPU usage is rather high in general. I have one VPN server and one VPN client connection setup (I had that before as well).

When I updated the FW I also setup asuswrt devicetracker in Hass.IO running on a device on my network. Not sure if that could be related.

How could I easiest debug this.
I need help to resolve it.

 

[martinr]

Which firmware were you running before you updated?

Have you tried disabling that device tracker you mentioned? What is it anyway; I haven’t come across it before?

When you say you get an authentication error, do you mean it asks for your wifi password?

“I can see that the CPU usage is rather high in general.” Is that higher than it used to be or have you only just looked at it?

Have you tried temporarily turrning off both OpenVPN services and your device tracker to see if that reduces the CPU level? And when you say high CPU level, what sort of numbers?



Have you looked in system log to see if there’s anything obvious?

Finally, can you remember at what point you last reset the router to its factory default state prior to inserting your custom settings?

I bet you’re sorry you asked, now!

 

[nihon]

I have managed to track it down I think. Both cores were peaking almost constantly at 100% CPU.
I disabled feature after feature and finally I removed the USB3 Hard-drive that I use as a server / AI disk.
Then CPU usage went back to < 5%.
(The disk was not accessed when I had the high CPU usage)

I cannot yet confirm 100% that was causing the hang. I need to wait and see now for a couple of days.
But it certainly resolved the extreme CPU usage (and extreme CPU usage I find likely to result in a stall sooner or later if the router does not catch up with the processing).

When the router crashed I could not even use the wired connection to get into the admin page.
I see no error logs that could tell me anything useful either.

So question now, has anyone else seen a problem with extreme high CPU usage when an USB disk is mounted?
Is it a known problem with a known solution?

 

[skeal]

Have you tried using "top" or "htop" to see what process has the system hung up?

 

[nihon]

@skeal, good point I did not check if also the SSH service went down. If not indeed one could login and use top or htop.
I have now removed all AI disk settings from the disk and will setup the usb disk again. If the problem occurs in a couple of days again I will try SSH.
But since I was not able to access the either of the routers admin page over HTPP or other LAN devices over HTTPS I am pretty confident the SSH would not have responded either when it hang.

 

[bbunge]

Check the USB drive for errors and correct if needed. Use the USB 2 port or switch the USB 3 port to USB 2 mode. Your router will be happier. Also add a thumb drive and add a swap file. Amtm is a great tool to do this.

Sent from my SM-T380 using Tapatalk

 

[nihon]

@martinr
Not sure what FW I had. Probably around 378 or 379. When the problems started I had upgraded to 384.8 then I tried the 384.9 with the same results when it came out.

I have not tried to disable the device tracker in hass.io I will do that if it shows up again after I have "reinstalled" the USB AI disk. Have a look at hass.io, it is a home assistant / automation software that I run on a raspberry-pi 3B to automate light switches and track who is home. AsusWRT is a plugin to it that makes the Asus router provide information to hass.io regarding what devices are connected and not. I run hass.io in a docker image on a raspbian OS (hass.io would normally be a complete OS otherwise but then one is locked out from modifying on the OS level).

I have never seen that high CPU usage before. A least on core was more or less always at 100% and manytimes both cores were about at 100% usage. See my other answer, it seems it was connected to the USB disk used as AI disk server.

I did not try earlier than today to turn services off. When I did so today the CPU usage did not went down until I removed the USB disk.
With the USB disk removed I turned on the services (VPN for example) but it did not increase CPU usage notable at least.

 

[nihon]

@bbunge
Okay but why having a USB3 connection if it does not work well?
I bought myself an USB3 drive just to make sure the router would not end up with bottlenecks and making sure fast access when streaming data.
I could switch to USB2 mode but would for sure prefer to keep USB3 mode. Is it a known issue with USB3 in the FW?

Can you please provide more details about the thumbdrive with a swap file?
I assume you mean that you recommend that I format the disk as EXT + Swap then? It is a ntfs file system at the moment.

 

[martinr]

@martinr
Not sure what FW I had. Probably around 378 or 379. When the problems started I had upgraded to 384.8 then I tried the 384.9 with the same results when it came out.

I have not tried to disable the device tracker in hass.io I will do that if it shows up again after I have "reinstalled" the USB AI disk. Have a look at hass.io, it is a home assistant / automation software that I run on a raspberry-pi 3B to automate light switches and track who is home. AsusWRT is a plugin to it that makes the Asus router provide information to hass.io regarding what devices are connected and not. I run hass.io in a docker image on a raspbian OS (hass.io would normally be a complete OS otherwise but then one is locked out from modifying on the OS level).

I have never seen that high CPU usage before. A least on core was more or less always at 100% and manytimes both cores were about at 100% usage. See my other answer, it seems it was connected to the USB disk used as AI disk server.

I did not try earlier than today to turn services off. When I did so today the CPU usage did not went down until I removed the USB disk.
With the USB disk removed I turned on the services (VPN for example) but it did not increase CPU usage notable at least.

“Not sure what FW I had. Probably around 378 or 379.”
Did you carry out a reset of the router to factory default settings AFTER flashing the new firmware followed by a MANUAL insertion of your custom settings (not from a backup)? Merlin mandated s factory reset between your old and new firmwares.

If you have not done this, don’t waste time doing anything else.

And use L&LD’s world-famous guude.
https://www.snbforums.com/threads/n...l-and-manual-configuration.27115/#post-205573

 

[nihon]

@martinr Good point I did not. I will do it.
Thanks for sharing the guide, what about the asuscomm.com dynamic dns.
Will I loose it?

 

[martinr]

@martinr Good point I did not. I will do it.
Thanks for sharing the guide, what about the asuscomm.com dynamic dns.
Will I loose it?

No, you won’t lose it. Of course, it will be blank after the factory reset, so you’ll simply re-insert it.

I’m sure this will resolve your problem, too. Let us know how you get on.