bortolino Posted March 6, 2018 Posted March 6, 2018 Hi, I downloadded the latest debian_stretch_next for OPZ+2 H5. Yes, I know it's experimental and not production-grade, I just want to report the issue. The problem is that apt-get (dpkg-deb) fails to decompress deb archives during the upgrade procedure. Tried it on a clean install multiple times on different microSDs. Are the packages really corrupted or is there something wrong with the image? Error log in the attachments P.S.: ubuntu_xenial_next randomly panics during apt update. h5.txt
teenytinycactus Posted March 7, 2018 Posted March 7, 2018 (edited) I got this error when I did update&upgrade on the same board, EXCEPT it only occurred for the kernel. I also got compression errors when I was pulling down the ROS2 git repositories. Are you using WiFi? Edited March 7, 2018 by teenytinycactus
bortolino Posted March 7, 2018 Author Posted March 7, 2018 10 hours ago, teenytinycactus said: I got this error when I did update&upgrade on the same board, EXCEPT it only occurred for the kernel. I also got compression errors when I was pulling down the ROS2 git repositories. Are you using WiFi? Yes, I'm using wifi. I found out that sometimes an apt clean + fix solves the issue, so I tested this script while true; do wget -q http://beta.armbian.com/pool/main/l/linux-4.14.0-rc7-sunxi64/linux-headers-dev-sunxi64_5.41.180305_arm64.deb -o out.deb md5sum out.deb sleep 3 done and I got a lot of mismatches. So there's something wrong with fs/networking. The microSD should be fine: works flawlessy on my RPi and passes H2testw. I'll try with another setup (different board/sd/network type) to exclude hw problems... UPDATE: I moved the SD to a new board, and problems disappeared! According to memtester I had a (very) faulty ram brick.
teenytinycactus Posted March 7, 2018 Posted March 7, 2018 5 hours ago, bortolino said: I moved the SD to a new board, and problems disappeared! According to memtester I had a (very) faulty ram brick. I am doing a memtest now and mine seems to be okay. Could you please advise on the arguments you used for memtester?
bortolino Posted March 7, 2018 Author Posted March 7, 2018 1 hour ago, teenytinycactus said: I am doing a memtest now and mine seems to be okay. Could you please advise on the arguments you used for memtester? sudo memtester 400 10 Could you try to run the hash script I posted earlier? On the faulty board I had inconsistent results.
teenytinycactus Posted March 7, 2018 Posted March 7, 2018 7 hours ago, bortolino said: sudo memtester 400 10 Could you try to run the hash script I posted earlier? On the faulty board I had inconsistent results. I ran also for 400 MB, and loop 1 was fine. I will leave memtester running for a few hours and see what happens. I did run the hash script for about 20 iterations - all the md5s were the same.
bortolino Posted March 8, 2018 Author Posted March 8, 2018 45 minutes ago, teenytinycactus said: I ran also for 400 MB, and loop 1 was fine. I will leave memtester running for a few hours and see what happens. I did run the hash script for about 20 iterations - all the md5s were the same. If RAM is fine it could be a microSD problem. Have you tried to change it? Have you got sporadic kernel panics?
teenytinycactus Posted March 8, 2018 Posted March 8, 2018 11 minutes ago, bortolino said: If RAM is fine it could be a microSD problem. Have you tried to change it? Have you got sporadic kernel panics? I am not sure if they are panics, but the device does hang fairly often when under stress (it has hung just now doing the memtest). I am currently using the serial interface provided via the microUSB (yes I understand this isn't a good idea and the problem might be PSU issues and/or voltage drop, I have ordered new parts and cables to rectify this.). If I connect via the UART0 serial, will I see the kernel panic output? I haven't tried another microSD as this is a brand new one from a brick and mortar store. It also worked flawlessly on the RPI Zero W. Also thanks for your help.
bortolino Posted March 8, 2018 Author Posted March 8, 2018 17 hours ago, teenytinycactus said: I am not sure if they are panics, but the device does hang fairly often when under stress (it has hung just now doing the memtest). I am currently using the serial interface provided via the microUSB (yes I understand this isn't a good idea and the problem might be PSU issues and/or voltage drop, I have ordered new parts and cables to rectify this.). If I connect via the UART0 serial, will I see the kernel panic output? I haven't tried another microSD as this is a brand new one from a brick and mortar store. It also worked flawlessly on the RPI Zero W. Also thanks for your help. Yes, you should see the panic message and it should reboot after a few seconds. If it just hangs it could be, as you said, a power problem.
Recommended Posts