I have a Rock64 4GB board, and love the capabilities of it. Unfortunately, I'm running into a problem that I'm hoping for help with.
I'm suddenly experiencing constant kernel panics. I've tried DietPi, Manjaro KDE, Manjaro XFCE, and OpenMediaVault--each on 2-4 different microSD cards, but all of them experience kernel panics between 10-90 minutes after booting. It's a very frustrating problem.
I've looked through the forums here, and tried the things I could find (like inspecting the board for soldering issues), but the board seems okay, as far as I can see. I purchased the board through an official vendor (AmeriDroid), and got the power adapter they recommended, so I don't think it's a power supply issue. I submitted a help ticket since I've tried all the troubleshooting steps I could find in the forums, but was told that I still need to post this here. Perhaps someone has some additional steps I can try?
If you have any advice or help you can give, I would be extremely grateful.
Thank you very much!
Reports of memory system failure are often seen.
I'm assuming that device cooling is sufficient,
I think you should check if the memory system is working properly.
If there is no problem with the memory system, it should at least pass the "memtester" test.
10-11-2021, 04:08 PM
(This post was last modified: 10-11-2021, 10:58 PM by clay.)
(10-11-2021, 01:24 PM)t4_4t Wrote: Reports of memory system failure are often seen.
I'm assuming that device cooling is sufficient,
I think you should check if the memory system is working properly.
If there is no problem with the memory system, it should at least pass the "memtester" test.
Thanks for the reply!
Cooling is definitely good. I have a pretty good sized heatsink on the CPU, and have the board in one of those open-sided, sandwich-style acrylic enclosures, with a fan blowing down on the CPU. I've watched the temp, and the crashes don't correlate to high CPU usage/high temp.
I ran memtester with three loops. An "echo $?" came back with a '0', so the memory appears to be okay.
(10-11-2021, 09:43 AM)clay Wrote: I have a Rock64 4GB board, and love the capabilities of it. Unfortunately, I'm running into a problem that I'm hoping for help with.
I'm suddenly experiencing constant kernel panics. I've tried DietPi, Manjaro KDE, Manjaro XFCE, and OpenMediaVault--each on 2-4 different microSD cards, but all of them experience kernel panics between 10-90 minutes after booting. It's a very frustrating problem.
I've looked through the forums here, and tried the things I could find (like inspecting the board for soldering issues), but the board seems okay, as far as I can see. I purchased the board through an official vendor (AmeriDroid), and got the power adapter they recommended, so I don't think it's a power supply issue. I submitted a help ticket since I've tried all the troubleshooting steps I could find in the forums, but was told that I still need to post this here. Perhaps someone has some additional steps I can try?
If you have any advice or help you can give, I would be extremely grateful.
Thank you very much!
Do you have any other devices attached to the board via USB2 or USB3. Are you using the board headless or with HDMI/Keyboard/Mouse. Maybe try the Armbian Distro as I've found it to be fairly stable. Do you have any log files you could inspect. I know you are using the official power supply but could it be overheating and causing your problem.
(10-11-2021, 06:49 PM)Rocklobster Wrote: (10-11-2021, 09:43 AM)clay Wrote: I have a Rock64 4GB board, and love the capabilities of it. Unfortunately, I'm running into a problem that I'm hoping for help with.
I'm suddenly experiencing constant kernel panics. I've tried DietPi, Manjaro KDE, Manjaro XFCE, and OpenMediaVault--each on 2-4 different microSD cards, but all of them experience kernel panics between 10-90 minutes after booting. It's a very frustrating problem.
I've looked through the forums here, and tried the things I could find (like inspecting the board for soldering issues), but the board seems okay, as far as I can see. I purchased the board through an official vendor (AmeriDroid), and got the power adapter they recommended, so I don't think it's a power supply issue. I submitted a help ticket since I've tried all the troubleshooting steps I could find in the forums, but was told that I still need to post this here. Perhaps someone has some additional steps I can try?
If you have any advice or help you can give, I would be extremely grateful.
Thank you very much!
Do you have any other devices attached to the board via USB2 or USB3. Are you using the board headless or with HDMI/Keyboard/Mouse. Maybe try the Armbian Distro as I've found it to be fairly stable. Do you have any log files you could inspect. I know you are using the official power supply but could it be overheating and causing your problem.
I have a wireless keyboard and wireless mouse (each with their own dongles) plugged into the USB 2.0 ports, and am running it connected to a monitor. But for what it's worth, I don't use the monitor much. I usually access the Rock64 remotely via SSH and RDP (when running a GUI).
I'm game to try Armbian, but it would be a bit disappointing to be limited to one OS, when others are more tailored to my use cases and interests. And it does seem that many Rock64 users use OMV and DietPi without issue (Manjaro, it seems, can be a bit hit & miss). But I'll give it a shot and see what happens.
As for overheating, it doesn't appear to be the issue. I have the Rock64 running in a case with tons of airflow, with a heatsink, and with a fan blowing directly onto the CPU/heatsink. I've also watched the CPU temp right up until the time of kernel panic a few times, and it's never when the CPU is getting hit hard. It seems truly random as far as I can see.
I'm totally up for providing any logs that could be helpful. Are there any that jump to mind as high potential candidates for highlighting issues?
Thanks a lot for the help!
From your report, the load seems light and cooling does not seem to be a problem.
However, looking at the symptoms alone, it seems to me to be similar to that of a bad memory (randomly crush occurring).
https://forum.pine64.org/showthread.php?tid=7387
It's a bit of an old report, but you could try lowering the memory clock as in this report and see if that improves the symptoms.
The only other thing I can think of is to connect a serial console and try to capture a crash report.
There might be something in there that could give you a clue.
Just a quick update on the Armbian test. I got it up and running, but after getting to the desktop, it only lasted about 15 minutes before crashing. So it doesn't appear to be OS-specific. 
Unfortunately, I also don't have easy access to buy or make a serial console cable. But it seems to me that we're honing in on this being some sort of hardware issue, not user error or software.
(10-12-2021, 03:07 AM)t4_4t Wrote: From your report, the load seems light and cooling does not seem to be a problem.
However, looking at the symptoms alone, it seems to me to be similar to that of a bad memory (randomly crush occurring).
https://forum.pine64.org/showthread.php?tid=7387
It's a bit of an old report, but you could try lowering the memory clock as in this report and see if that improves the symptoms.
The only other thing I can think of is to connect a serial console and try to capture a crash report.
There might be something in there that could give you a clue.
Just runs memtester overnite for system memory testing.
(10-12-2021, 12:51 PM)clay Wrote: Just a quick update on the Armbian test. I got it up and running, but after getting to the desktop, it only lasted about 15 minutes before crashing. So it doesn't appear to be OS-specific.
Unfortunately, I also don't have easy access to buy or make a serial console cable. But it seems to me that we're honing in on this being some sort of hardware issue, not user error or software.
try different 5V 3A power supply if you have them handy and don't attached any USB device at the Rock64 USB ports.
I booted into a fresh Armbian install and ran the board with no USB devices attached. I was optimistic at first, because it ran overnight (around 10 hours). But then it crashed. I tried bringing it back up multiple times, but after that first crash, it only lasted about 5-15 minutes each time before it crashed again. I'm not sure if it was just coincidence that it ran so long at first, or if there is, indeed, something related to USB.
Unfortunately, all the other 5V/3A power supplies in my collection have micro USB connectors, not the barrel connector needed for the Rock64.
Thanks again for the help troubleshooting!
|
