RT-AC66u reboots under heavy load scenario

[xlvii]

Hello!
I'm using Asus RT-AC66u running latest 376.49_5 Merlin's firmware.
Let me describe configuration of the system and then the load scenario which crashes RT-AC66u software (or hardware?).
The Internet connection is so-called 'Russian PPTP dual-access', which means ISP has its local network between users (for sharing service like emule or FTP) and connection to the Internet is done via VPN PPTP connection. Routing to ISP's local network addresses is achieved by one static routing rule (MAN). Internet speed is 60Mbit/s DL/UL, speed in ISP's local network is 100Mbit/s DL/UL.
USB HDD is connected to RT-AC66u (WD My Passport Ultra 2 Tb, formatted in NTFS). Download Master is installed and Samba sharing configured.
Next, two devices:
1) Laptop: Samsung R560, connected via 1Gbit Ethernet adapter
2) Tablet: Samsung Nexus 10, connected via 5Ghz 802.11n Wi-Fi connection.
Now, the scenario.
1) On the laptop I start downloading any huge file from one of the ISP's local network FTP servers directly to network shared folder on the USB-HDD, connected to RT-AC66u. The speed gets up to 11000 KB/s (but flapping).
2) During the operation, I run Speedtest app on the Tablet. In first seconds FTP download hangs and Speedtest hangs (appeared to be not really needed)
...and then after 30-120 seconds router reboots itself.
I've performed scenario several times, the result is always the same.
Router has been working fine for me for several months now, never encountered self-reboots before.
Any help or suggestions would be appreciated!
P.S. Sorry for my English.

 

[huotg01]

When doing that, is it possible for you to keep an eye on System Status graphs on the home page and see what is going on with the memory and the CPU ? The idea is to see if you run out of memory.

 

[Hexabion]

Have you tried disabling Download Master.

I would install entware and run transmission instead, or the equivelent for arm (if AC66u is arm?).

11000KBs = roughly 10.7MBs = which is roughly 85Mbits speed, meaning you're getting good speed, and probarbly bottling up the LAN>WAN or hanging the cpu, meaning you should set a limit, thus use transmission instead.

Then again, this could be the FW not playing well with your setup, have you tried 49._2 or 49._4?

I'm using N66u though, but I use 49._2, whichs seems very stable. I seldom use the very last version of anything as its often stability vs "being first" - i prefer stability over the ladder.

Also, open a ssh and run "top"

 

[Gravityz]

my router was also renbooting everyday
i thought it was normal(which is not)

go through every setting(especially the usb section) and disable everything you do not use/need

eg mine has settings under 3g/4G where it said USB modem YES

if you do not have an usb modem with simcard this one needs to be turned of.

after that reboot and see if it helps.

 

[xlvii]

Ok, thanks everyone for the answers so far!
I was able to find out that running Speedtest on the tablet is not even necessary for reboot, it just speeds it up.
huotg01,
I watched for memory while downloading file. At the beggining free mem amount quickly drops from 100mb to ~32mb, but then slowly gets to ~65mb at the moment of reboot. You can see it on the screenshot below.
https://www.dropbox.com/s/g1os673bhatq939/asus.png?dl=0
Also I noticed that CPU load drops to zero just the second before reboot.
Didn't find anything in the log file on JFFS partition at the time of reboot.
Hexabion,
Download Master isn't actually working in the background (no any tasks, whatsoever). Router handles load fine when I download the file from FTP to my local laptop HDD instead of USB-HDD connected to router. The CPU load in this case doesn't go above 80% and free mem is 100mb constant.
I've run the 'top' command through telnet during the experiment. Top load process is /usr/bin/smbd. It loads CPU from 12% to 85%. Screenshot below.
https://www.dropbox.com/s/f9kn4xxo8dlaf26/asus2.png?dl=0

 

[huotg01]

Ok, thanks everyone for the answers so far!
I was able to find out that running Speedtest on the tablet is not even necessary for reboot, it just speeds it up.
huotg01,
I watched for memory while downloading file. At the beggining free mem amount quickly drops from 100mb to ~32mb, but then slowly gets to ~65mb at the moment of reboot. You can see it on the screenshot below.
https://www.dropbox.com/s/g1os673bhatq939/asus.png?dl=0
Also I noticed that CPU load drops to zero just the second before reboot.
Didn't find anything in the log file on JFFS partition at the time of reboot.
Hexabion,
Download Master isn't actually working in the background (no any tasks, whatsoever). Router handles load fine when I download the file from FTP to my local laptop HDD instead of USB-HDD connected to router. The CPU load in this case doesn't go above 80% and free mem is 100mb constant.
I've run the 'top' command through telnet during the experiment. Top load process is /usr/bin/smbd. It loads CPU from 12% to 85%. Screenshot below.
https://www.dropbox.com/s/f9kn4xxo8dlaf26/asus2.png?dl=0

I don't know....If the device reboots following a crash, it is quite possible that the real condition that makes it crash is not logged.
Just a few more ideas to explore:
- trying the other usb port
- trying an external powered usb disk (I think that the Passport is powered by the usb port, isn't ?
- adding a swap file...(I should say that it doesn't look required, but...)
- doing a jump back to a previous firmware (.43 ?)

 

[xlvii]

OK, considering all your advices, I think that the optimal plan is:
1) Check if router reboots if uploading from laptop's local HDD instead of remote FTP, monitor RAM and CPU load. If it won't reboot, the problem is more likely to be software one, than hardware.
2) Try to connect USB-HDD with some external power. Also try different USB-HDD.
3) Rollback firmware to .43

As soon as I get these things done, I will inform you about results.

 

[xlvii]

Ok, so I tried the following:
1) Flashing Merlin's 374.43 firmware
2) Flashing official Asus firmware 3.0.0.4.376.3754
In both cases I got the problem showed itself, though reboots were often replaced with system hangs (internet working, but can't access Router's Web Page or even connect with Telnet).
3) Using USB Flash drive instead of USB-HDD. Same story. BTW, no Download Master was installed in this case, so it can't be the cause.
4) Uploading from laptop's local HDD instead of remote FTP - it woked perfectly fine! The speed was stable and constant at around 8000 KBytes/s and 4GB file uploaded to Router's USB-HDD nicely without any hangs or reboots.
5) Downloading file from FTP using Router's Download Master. And it worked well too! Speed was about 8000KBytes/s too.
6) Forcibly reduce speed of download in Total Commander. I reduced it to 1500KBytes/s, and download lasted a little longer (about 1500MB vs usual 300-700MB before hang/reboot).

So, summing up the results:
-Downloading from FTP to Router's USB-HDD using Router's Download Master: constant speed 8000KBytes/s, no problems whatsoever
-Downloading from FTP to laptop: constant speed 10000 KBytes/s, no problems whatsoever
-Uploading from laptop to Router's USB-HDD (or USB flash): constant speed 8000KBytes/s, no problems whatsoever
-Downloading from FTP to Router's USB-HDD using laptop (Windows 7 x86, Total Commander): speed flapping very intensively from 1000 to 10000 KBytes/s, download interrupts for 1-2 seconds every 30-60 seconds, download lasts about 1-2 minutes and then router hangs or reboots itself.

I'm running totally out of ideas at this point... I'm going to try another laptop, because there's one thing that confuses me: laptop's Ethernet adapter supports 1GBit/s, and Router does too, but Windows reports that speed connection is 100Mbit/s even if I forcibly set Ethernet speed to 1Gbit/s in adapter's properties (in Device Manager). Cable is fully (eight lines) wired, the stock noodle cable from AC66u BOX.
So I'm going to test scenario on different laptop.
If you guys have any more ideas of what to try or to do for further investigation, you're very welcome to share...

 

[huotg01]

Not a funny trip...

No other ideas; will just repeat 2 already mentioned:
- trying the other usb port
- using a swap file

Nothing makes me think that it would solve the problem, but if it is not too difficult for you to make these two tests ?

(By the way, are you always using the same file and the same FTP server ? Have you tried a huge file from another FTP server, and another huge file from the same FTP server?)

 

[Hexabion]

-Downloading from FTP to Router's USB-HDD using laptop (Windows 7 x86, Total Commander): speed flapping very intensively from 1000 to 10000 KBytes/s, download interrupts for 1-2 seconds every 30-60 seconds, download lasts about 1-2 minutes and then router hangs or reboots itself.

What you're actually doing here is two actions and using double bandwidth:

A: File transfer internet-> (using inet bandwidth) -> router-> (using LAN/eth/wifi bandwidth) -> laptop

while youre also simultaniously:

B: File transfer laptop -> (using even more LAN/eth/wifi bandwidth) -> routers USB HDD.

B depends on the FTP-client software, some FTP-clients support site-to-site directly, meaning it doesnt buffer via your laptop, but some clients to, if your client does the ladder, then your'e bogging up the bandwidth in all directions, allthough it shouldnt "crash" a router though.

Are you by chance using the inbuilt minidlna then? Cause on my router, the inbuilt minidlna goes completely bonkers when I do fileoperations towards the routers USB HDD.

Have you tried turning on/off QOS, NAT-acceleration and Spanning Tree (under LAN), have you tried an alternative ftp-server (by disabling the one in the GUI, and install one in optware/debian instead on the router?)
And/or tried setting QOS to someting like a couple of Mb's below your ISP's cap? (just to pin out if its bw-strangling thats killing the router)

Nonetheless, I really hope your pinpoint your problem and get to fix it, please share with us if you find enlightenment.