Randoms reboot NanoPi M4V2

[i5Js]

Hello all, 

 

I have this board running in a k3s kubernetes cluster, and almost everyday, different time, the board reboots. Please, could you help me to figure out which is the problem? The armbianmonitor output is bellow:

http://ix.it/2OOZ

 

Many Thanks,

[Werner]

54 minutes ago, i5Js said:

http://ix.it/2OOZ

Invalid link

[i5Js]

Thanks Werner, I've uploaded it again:

 

http://ix.io/2PI6
 

[Igor]

1 hour ago, i5Js said:

and almost everyday, different time, the board reboots.

Unstable / insufficient powering would be 1st guess.

[i5Js]

8 minutes ago, Igor said:

Unstable / insufficient powering would be 1st guess.

Interesting, because I'm using the official power adapter from friendelec

[Igor]

10 minutes ago, i5Js said:

Interesting, because I'm using the official power adapter from friendelec

5V powering via USB-C is on the edge (especially for hungry hw) if it comes via official or unofficial power adapter / cable.

[i5Js]

21 minutes ago, Igor said:

5V powering via USB-C is on the edge (especially for hungry hw) if it comes via official or unofficial power adapter / cable.

mmm perhaps the cable is the problem... how could I check it? Also I'm using a an adapter to fit from Asia to Europe

 

Anyway, is there any other alternatives you can recommend me?

[Werner]

Try different PSU/cable. Power via GPIO

Measure voltages from PSU output and on the board directly to get an idea about the voltage drop across all parasitic resistances.

[i5Js]

Thanks both, I’ll start replacing the cable.

[i5Js]

The cable made something different, since the board was up almost 1.5 days, but it was rebooted again. 

 

@Werner I'm very noob, can you explain how can I power it up from GPIO?  or which other PSU could work? I read in the internet some guys are using Raspberry's Pi 4 PSU, but reading the specs, are lower than the official from friendelec.

 

Many thanks.

[Werner]

According to their wiki page https://wiki.friendlyarm.com/wiki/index.php/NanoPi_M4V2

put 5v and ground to pin 2 and 6.

 

49 minutes ago, i5Js said:

Raspberry's Pi 4 PSU

They come with slight overvoltage of 5.3volts afaik. Compensates for losses through connectors and cables. Worth a try. However any PSU featuring output slightly above 5V will do.

[NicoD]

I've done a lot of tests with the M4V2 and powering and even at 4V it stayed stable. I've never had a crash running the legacy kernel. 
Best thing to check voltage is to use a usb voltage meter in the USB ports of the SBC. Also at the input and see if the difference is big. Using good cables is necessary. Even most USB-C cables are not made for power. This SBC can use up to 2A, most cheap USB cables can't handle that. 
A Raspberry Pi crashes quickly, and is very unstable with even 4.9V. But not the M4V2.

What is connected to it? Maybe a USB over voltage? You could also try a legacy kernel image to see if it's a kernel issue. Also using uart connected to another device you'd probably be able to see what went wrong. 

[i5Js]

Thanks all for the messages, I've tried with another one, 6.5V, and it's the same, it was rebooted 3.5hours ago. I don't know if it could be a kernel problem, I'm using also a M4V1 with SATA hat, and kernel 5 and works flawless (Up time:       27 days 1:15), so I think it is related to the PSU/cable...

My main problem here is I'm not good with electric stuff, meters etc, I have an standard meter, but I don't know even how to use it

 

Sorry for asking @NicoD with the usb meter, should be plug and see?  if so, I think even I could make it.

 

I'm going to look for a good usb cable... 

 

Thanks.

[i5Js]

The new cable arrived. It supports up to 4A so fingers cross.

[i5Js]

Well, the cable is not the problem, the board rebooted 36 mins ago.... so frustrating.

[Lennyz1988]

You could also

10 minutes ago, i5Js said:

Well, the cable is not the problem, the board rebooted 36 mins ago.... so frustrating.

 

I used to have a lot of hangs with older mainline kernels. Since I am on the latest everything is running smoothly for me. You could try the latest kernel. You can't download this kernel from the website. You can download this kernel through armbian-config.

 

Armbian 21.02.2 Buster with Linux 5.10.16-rockchip64

[i5Js]

Updated to the latest... let's see.

[piter75]

6 hours ago, i5Js said:

Updated to the latest... let's see.

If it fails which I think it will... Can you try this image? https://users.armbian.com/piter75/Armbian_21.05.0-trunk_Nanopim4v2_buster_current_5.10.18_minimal-stability-fix.img.xz

It contains a fix/workaround to dvfs issues I found and makes all of my M4V2 units running stable. Tested with hours of memtester running without failures. It failed pretty quick without the fix.

[Glock24]

Hello,

 

I recently got a Nanopi M4v2 on which I installed Armbian Buster current and updated the system, including kernel.

 

After a few hours the board would crash, requiring a power cycle.

 

Aftet reading this thread I decided to try a fixed frequency, so I left it at 2GHz. With a fixed clockspeed I ran stress tests, several runs of sbc-bench, browser benchmarks,  used it to open documents in Libreoffice, watched videos using the browser, etc. and it was totally stable for over 5 days straight.

 

I only had problems with Chromium crashing, but Firefox (or equivalent) worked fine.

 

I changed the max and min clocks to defaults and it locked up about 4 hours later.

 

The board was working headlessly and I accessed it remotely using xrdp or dwservice.net.

 

I don't have it connected to UART at the moment because I don't have another computer nearby to connect it to where the board is placed, but if it helps I can find a way to connect it and give you the output.

[i5Js]

Hi @Glock24 thanls for your message, could you please share your frequency settings? Did you set with armbian-config right?

[piter75]

@i5Js any constant frequency should do, switching cpufreq to performance profile also solved M4V2 stability issues in the past

 

@i5Js @Glock24 I would appreciate if you verified the image referred in the previous post. If it works without crashes and we merge the change into master we could finally have stable M4V2 in Armbian current out of the box.

 

[i5Js]

Hi @piter75 Thanks! I'll check it, and if new reboots comes, I'll try that kernel. I've not changed any constant yet.

[i5Js]

Record!! Up time:       1 day 2:26

 

Update: Up time:       1 day 16:43 

[Lennyz1988]

14 hours ago, i5Js said:

Record!! Up time:       1 day 2:26

 

Update: Up time:       1 day 16:43 

 

Just to confirm you are on this kernel right? Linux 5.10.16-rockchip64

[i5Js]

6 hours ago, Lennyz1988 said:

 

Just to confirm you are on this kernel right? Linux 5.10.16-rockchip64

 

Yes!

 

Welcome to Armbian 21.02.2 Focal with Linux 5.10.16-rockchip64

 

System load: 10%           Up time:       1 day 23:13

[i5Js]

Time to test 5.10.18 image... it just rebooted. 2 days and a half up.

@piter75 Sorry for asking, how can install the image without reinstall the hole system?

[piter75]

@i5Js you can download these packages and install them with apt/dpkg: https://users.armbian.com/piter75/nanopim4v2-stability-fix/

You need both of them for the fix.

 

When they are correctly installed and the board is rebooted you should see this message in your dmesg:

piter@nanopim4v2-4:~$ dmesg | grep rk808-regulator.*buck
[    2.840331] rk808-regulator rk808-regulator: max buck steps per change: 4

The last "4" means you have the fix. No message or "8" means you don't have a fix.

[i5Js]

Hi @piter75, I've just applied the patch, and rebooted the pi

 

i5js@noldork3sN2:~$ dmesg | grep rk808-regulator.*buck
[    2.883990] rk808-regulator rk808-regulator: max buck steps per change: 4

 

I'll keep you posted.

 

Thanks!

[piter75]

On 3/7/2021 at 10:50 AM, i5Js said:

I'll keep you posted.

Great. Looking forward to hear about your results.

[i5Js]

Up time:       2 days 8:57 and counting  

 

Looks promising.