Heimdall Posted August 8, 2018 Posted August 8, 2018 Hi guys, After some time I get enough brave to do apt upgrade on my Odroid C2 (after month or two).. I updated following packages: Start-Date: 2018-08-07 11:09:56 Commandline: apt upgrade Install: html2text:arm64 (1.3.2a-18, automatic), debconf-utils:arm64 (1.5.56+deb8u1, automatic) Upgrade: ncurses-term:arm64 (5.9+20140913-1+deb8u2, 5.9+20140913-1+deb8u3), libtinfo5:arm64 (5.9+20140913-1+deb8u2, 5.9+20140913-1+deb8u3), linux-dtb-odroidc2:arm64 (5.40, 5.54), libncurses5:arm64 (5.9+20140913-1+deb8u2, 5.9+20140913-1+deb8u3), libncursesw5:arm64 (5.9+20140913-1+deb8u2, 5.9+20140913-1+deb8u3), ncurses-bin:arm64 (5.9+20140913-1+deb8u2, 5.9+20140913-1+deb8u3), base-files:arm64 (8+deb8u10, 8+deb8u11), patch:arm64 (2.7.5-1, 2.7.5-1+deb8u1), libldap-2.4-2:arm64 (2.4.40+dfsg-1+deb8u3, 2.4.40+dfsg-1+deb8u4), linux-image-odroidc2:arm64 (5.40, 5.54), armbian-config:arm64 (5.40, 5.54), linux-u-boot-odroidc2-default:arm64 (5.40, 5.49), docker-ce:arm64 (18.05.0~ce~2.1.rc1-0~debian, 18.06.0~ce~2.1.rc1-0~debian), armbian-tools-jessie:arm64 (5.40, 5.44), tzdata:arm64 (2018d-0+deb8u1, 2018e-0+deb8u1), armbian-firmware:arm64 (5.40, 5.45), ncurses-base:arm64 (5.9+20140913-1+deb8u2, 5.9+20140913-1+deb8u3) End-Date: 2018-08-07 11:10:45 After upgrade I am having freeze almost every 12 hours - not able to connect to the odroid via SSH (running headless) and I have to replug power (I am using official power adapter from Odroid via jack). Here is link to logs etc. any ideas what it could be? I will try to set up nmon or zabbix to monitor health of mem / cpu before crash.
Heimdall Posted August 8, 2018 Author Posted August 8, 2018 It freeze again, I tried to plug monitor to see something.. but no output at all :-/ going to set up nmon
Heimdall Posted August 9, 2018 Author Posted August 9, 2018 It stuck again. Here is the graphs from nmon - https://imgur.com/a/K7T4N3 Nothing special to see actually (all looks normal to me..). Will try to find out if I can go back with kernel or firmware.
Igor Posted August 9, 2018 Posted August 9, 2018 On 8/8/2018 at 3:52 PM, Heimdall said: After upgrade I am having freeze almost every 12 hours We proceeded to 3.16.y kernel as did the Hardkernel. Nothing interesting in the kernel commit logs: https://github.com/hardkernel/linux/commits/odroidc2-v3.16.y our build is very recent so it is virtually the same. Can you check 4.14.y/4.17.y kernels? For that, you need to change u-boot and boot scripts. They are getting to that state, where we could just forget about old legacy kernels. Soon. What is your use case?
Heimdall Posted August 9, 2018 Author Posted August 9, 2018 Is there any howto for upgrade? My use case is just using few docker images (5 running). OS is running on eMMC card and data are on USB disk.
Heimdall Posted August 11, 2018 Author Posted August 11, 2018 I found option to upgrade kernel from armbian-config, but after upgrade to next kernel, after reboot is just green screen on HDMI and nothing else.. going to try clean next installation..
Igor Posted August 11, 2018 Posted August 11, 2018 47 minutes ago, Heimdall said: I found option to upgrade kernel from armbian-config, but after upgrade to next kernel, after reboot is just green screen on HDMI and nothing else.. going to try clean next installation.. This problem was fixed with this commit but haven't got its way to the repository yet.
Heimdall Posted August 11, 2018 Author Posted August 11, 2018 I just installed clean image (its not a big deal, since data are on external USB disc), but I am not able now to get disk even detected in system (its not even blinking), but this was working correctly on 3.16 kernel.. any ideas? I created logs where I can see, that USB hub is detected, but when I put USB disk on, nothing (in dmesg or messages) - http://ix.io/1jSw Not even keyboard + mouse which I have in powered USB hub is loaded.. not a single line in dmesg after connection (tested all 4 ports). Don´t know if this can be named as "They are getting to that state, where we could just forget about old legacy kernels." :-)
Heimdall Posted August 11, 2018 Author Posted August 11, 2018 Just now, Igor said: This problem was fixed with this commit but haven't got its way to the repository yet. Heh, this kind of commits should be in repo asap
Igor Posted August 11, 2018 Posted August 11, 2018 10 minutes ago, Heimdall said: Heh, this kind of commits should be in repo asap Getting to the main repository needs general testing. It's not the only thing that has been changed. Until we are running this project on a hobby level ... it has to wait. Barbecue has higher priority
Heimdall Posted August 11, 2018 Author Posted August 11, 2018 Just now, Igor said: Getting to the main repository needs general testing. It's not the only thing that has been changed. Until we are running this project on a hobby level ... it has to wait. Barbecue has higher priority And any ideas about USB problems in following post?
Igor Posted August 11, 2018 Posted August 11, 2018 USB generally works but sometimes hotplugging doesn't work. Reboot helps and latest changes regarding USB are probably not in the stable repo ... they are for sure in beta/nightly.
Heimdall Posted August 11, 2018 Author Posted August 11, 2018 16 minutes ago, Igor said: USB generally works but sometimes hotplugging doesn't work. Reboot helps and latest changes regarding USB are probably not in the stable repo ... they are for sure in beta/nightly. Nope, reboot was first what I was trying.. mkey, back to 3.16 and I will write to my forehead "ARM boards are not worth it.. x86 ftw" (but I am glad what you and whole Armbian team doing, but support from ARM producers is shitty unfortunately).
Igor Posted August 11, 2018 Posted August 11, 2018 52 minutes ago, Heimdall said: Nope, reboot was first what I was trying.. mkey, back to 3.16 and I will write to my forehead "ARM boards are not worth it.. x86 ftw" (but I am glad what you and whole Armbian team doing, but support from ARM producers is shitty unfortunately). 6 Comparing experimental mainline support on certain cheap ARM hardware vs. mainstream x86 is pointless. ARM boards are chosen by a use case in first place. If you want a modern kernel, certain things will simply not work. USB must work and it should. This (and usually network) is a minimum level that we start to provide test support. For certain uses cases, this is already good enough. Since ATM I can't check if the latest update actually broke USB I can't argue about. USB used to work on a modern kernel with a few limitations ... ARM corp. only creates a design which is licensed to chip makers. Chip on C2 was made in Amlogic corp., which helps the community develop mainline kernel support via this project http://linux-meson.com/doku.php I hope they(Amlogic) are not underestimating the scale of this process ... which takes months to years per chip family. We are more on the implementation, testing and maintenance side in this process.
Igor Posted August 11, 2018 Posted August 11, 2018 I made a few tests and you can see below that USB is working all the time. [ 3.810953] scsi 0:0:0:0: Direct-Access Multiple Card Reader 1.00 PQ: 0 ANSI: 0 [ 4.015361] random: crng init done [ 4.015376] random: 7 urandom warning(s) missed due to ratelimiting [ 4.366231] sd 0:0:0:0: [sda] 15269888 512-byte logical blocks: (7.82 GB/7.28 GiB) [ 4.367109] sd 0:0:0:0: [sda] Write Protect is off [ 4.367148] sd 0:0:0:0: [sda] Mode Sense: 03 00 00 00 [ 4.367938] sd 0:0:0:0: [sda] No Caching mode page found [ 4.367962] sd 0:0:0:0: [sda] Assuming drive cache: write through [ 4.374477] sda: sda1 Nightly builds upgrade testing: Odroid C2http://ix.io/1jU0 Odroid XU4, started with oldest Ubuntu Xenial, kernel 3.10.y -> NEXT -> DEV -> DEFAULT -> NEXThttp://ix.io/1jU1 XU4 ran from eMMC I found few problems which needs to be fixed before sending out a new update ...
Heimdall Posted August 14, 2018 Author Posted August 14, 2018 On 8/11/2018 at 12:54 PM, Igor said: Comparing experimental mainline support on certain cheap ARM hardware vs. mainstream x86 is pointless. ARM boards are chosen by a use case in first place. If you want a modern kernel, certain things will simply not work. USB must work and it should. This (and usually network) is a minimum level that we start to provide test support. For certain uses cases, this is already good enough. Since ATM I can't check if the latest update actually broke USB I can't argue about. USB used to work on a modern kernel with a few limitations ... ARM corp. only creates a design which is licensed to chip makers. Chip on C2 was made in Amlogic corp., which helps the community develop mainline kernel support via this project http://linux-meson.com/doku.php I hope they(Amlogic) are not underestimating the scale of this process ... which takes months to years per chip family. We are more on the implementation, testing and maintenance side in this process. That´s for long discussion, but computer (some intel nuc like) for 130€ will be in the end better for me (if I count spent hours with problems) then Odroid for 100€ But when I was buying Odroid at release, I was hoping for great support, but actually, Hardkernel doesnt care too much about C2. It suprise me that about XU4/MC1/HC1/HC2 they care much more And I want just headless server. I am not talking about some ARM for Kodi, because that is also not so well developed.. so for headleass.. do I want too much? On 8/11/2018 at 6:52 PM, Igor said: I made a few tests and you can see below that USB is working all the time. [ 3.810953] scsi 0:0:0:0: Direct-Access Multiple Card Reader 1.00 PQ: 0 ANSI: 0 [ 4.015361] random: crng init done [ 4.015376] random: 7 urandom warning(s) missed due to ratelimiting [ 4.366231] sd 0:0:0:0: [sda] 15269888 512-byte logical blocks: (7.82 GB/7.28 GiB) [ 4.367109] sd 0:0:0:0: [sda] Write Protect is off [ 4.367148] sd 0:0:0:0: [sda] Mode Sense: 03 00 00 00 [ 4.367938] sd 0:0:0:0: [sda] No Caching mode page found [ 4.367962] sd 0:0:0:0: [sda] Assuming drive cache: write through [ 4.374477] sda: sda1 Nightly builds upgrade testing: Odroid C2http://ix.io/1jU0 Odroid XU4, started with oldest Ubuntu Xenial, kernel 3.10.y -> NEXT -> DEV -> DEFAULT -> NEXThttp://ix.io/1jU1 XU4 ran from eMMC I found few problems which needs to be fixed before sending out a new update ... Interesting, for me it was not working on clean NEXT installation. With default (3.16) kernel it works well.. with clean installation also system is stable now and not crashing every 12 hours
Recommended Posts