animous Posted July 19, 2017 Posted July 19, 2017 (edited) Running on Virtualbox-compiled Armbian_5.32_Orangepiplus_Ubuntu_xenial_dev_4.11.5.img. I installed docker and am trying to run a pihole container. Because it's a new image, it tries to download it. But it never gets to the end, because it always freezes. OrangePi+ stops answering pings and I have to hard reboot. Here's almost what it looks like before locking up. I've also had it lock up during file transfers, differs from one time to the other. Spoiler Unable to find image 'diginc/pi-hole:arm' locally arm: Pulling from diginc/pi-hole 75419788cca0: Retrying in 5 seconds a270c0806eb4: Download complete 2a3c870a1902: Download complete b3d4c2f8bb93: Download complete fada51ac62da: Verifying Checksum ccaf95dfea4a: Download complete a62a109564e3: Download complete 693c0685d05f: Download complete 8ce104f7cdb9: Download complete 3734cf704ceb: Download complete 9d19776e160c: Download complete 6416ceedd127: Download complete 3f6964657b20: Download complete 9a5c748a3efe: Retrying in 5 seconds 98afe4f8c448: Download complete e0fb1e79297f: Download complete 93cd165c9105: Download complete 8d7f2a26f4f6: Download complete a7af1024353f: Download complete I have a Samsung Evo+ Class 10 32GB. I have tried different 5V/2.4A power supplies. I quickly checked logs in /var/log, but crash, stack trace or anything. Do I have to manually apply the 0001-sunxi-h3-Fix-PLL1-setup-to-never-use-dividers.patch patch from megous/linux when building my image? Any pointers on how to debug and get rid of these lockups would be greatly appreciated. Thanks! Edited July 19, 2017 by StuxNet Strictly screen real estate.
StuxNet Posted July 19, 2017 Posted July 19, 2017 You seem to know what you're doing since you've provided PS and SD card verification.... but I've gotta ask. How are you writing the kernel to the SD card? I don't mind being the bad guy here and it's the only newb thing you don't specify.
Igor Posted July 19, 2017 Posted July 19, 2017 3 hours ago, animous said: But it never gets to the end, because it always freezes. Wireless or Ethernet connection? IIRC there are some problems with wireless and well, network driver will also be replaced if not already?
animous Posted July 19, 2017 Author Posted July 19, 2017 @Stuxnet: The kernel is getting written as part of the whole image file using Etcher on OSX. @Igor: Wired. Obviously I doubt the CPU is getting used much during that download, but just in case I ran some stress tests (--cpu 4 --vm 1 --io 1 --hdd 1), to eliminate the variable of thermal regulation because of the CPU overheating. All pass. Spoiler 23:12:38: 1296MHz 2.81 3% 0% 1% 0% 0% 0% 61.2°C 23:12:44: 1296MHz 3.22 3% 0% 2% 0% 0% 0% 63.6°C 23:12:49: 1248MHz 3.61 3% 0% 2% 0% 0% 0% 65.1°C Time CPU load %cpu %sys %usr %nice %io %irq CPU 23:12:54: 1248MHz 4.04 3% 0% 2% 0% 0% 0% 65.7°C 23:13:00: 1248MHz 4.44 3% 0% 2% 0% 0% 0% 66.5°C 23:13:11: 480MHz 4.43 3% 0% 2% 0% 0% 0% 51.1°C 23:13:16: 480MHz 4.08 3% 0% 2% 0% 0% 0% 47.1°C^C
Igor Posted July 19, 2017 Posted July 19, 2017 Network driver is buggy and issues could manifest in cases when lot's of traffic is involved. Perhaps. Try to rule out onboard wired network, by using some USB based.
animous Posted July 19, 2017 Author Posted July 19, 2017 Bingo. Worked with USB ethernet! Thanks!!! Now waiting for my serial to usb cable to arrive from China to better understand why my orangepi won't boot when I compile the same .img, but with F2FS instead of ext4. Most likely I'll be posting again in 3-4 weeks. 2
chwe Posted July 25, 2017 Posted July 25, 2017 On 20.7.2017 at 1:46 AM, animous said: Now waiting for my serial to usb cable to arrive from China With an el cheapo CHG340 serial to USB or is it claimed to be a FTDI? I have three of them lying around. The CHG340 works just fine. Both FTDIs are fake, one works fine the other one made me crazy. So test your USB to serial before you run into 'unexpected issues on serial output'. Even if FTDI decided to no longer brick fake FTDIs with their driver. 1
Recommended Posts