• Posts

  • Joined

  • Last visited

Recent Profile Visitors

The recent visitors block is disabled and is not being shown to other users.

berobispo's Achievements

  1. Hi! my odroid HC2 is running pretty great on armbian 5.90 for >1,5 years (24/7) now. Running a nextcloud instance, with decent perfomance, even when clocked to 600/1200 Mhz in the summer, because of one (1) crash, that I think might have been heat-related. The thing I'm wondering about: in htop, it shows only ( .LITTLE) cores for eight lines, but the cpu is a (.big.LITTLE) architecture and my other HC2 sbc, which runs armbian 20.02.7, show "(.big.LITTLE)" in htop. Perfomance in sysbench is equal on both boards, so this might mean nothing, but it still seemed strange to me. Anyone knows why htop only shows "( .LITTLE)" cores?
  2. ahh okay. The reason I even realized that my HC2 is not running the most recent armbian is that I have another one, which I set up some months later and this other one shows the CPU cores differently in htop: "(.big.LITTLE)", whereas the HC2 running the nextcloud only shows 8 lines of "( .LITTLE)", as if there were no .big cores. When comparing those two machines, which I thought were running on the same OS, I found out that in the months between setting up these two machines, there must have been the armbian OS update. So, I guess I'll stay with stretch (and try to figure out why there are only .LITTLE cores showing). Thanks!
  3. Thanks! I think the kernel stays the same in the new image, but since debian is upgraded, I should start from a fresh image? But is there actually a downside by staying on stretch? Security-wise I guess the answer is: always stay up-to-date, right?
  4. Hi! I've been running two sbc (Odroid HC2) for about 1,5 years. I recently realised that the os it is running on is outdated (Debian Stretch with Armbian Linux 4.14.150-odroidxu4). There is a nextcloud instance running on it and I really (!) like the stability of my setup. So far, I've had one (1) crash in 1,5 years on a system that is running 24/7, maybe heat related in the hot summer. Reboot automatically once a week, nightly off-site backup, login via ssh-keys. So my question is...: Shall I update and if yes shall I do this from a fresh install or is there a built-in way? Thanks!
  5. Hi! my nextcloud instance on an odroid hc2 running armbian has been running stable and smoothly for several months now. Thanks Igor and the whole armbian team! I have one question though: Task: I convert videos with ffmpeg to reduce file size and optimize them for streaming observed behaviour: the CPU is getting very hot! In the evening up to 100 °C (see below), but at noon I have seen temperatures of up to 115 °C! I read somewhere that 110 °C is the critical temperature for the board and the stock hardkernel image has an emergency shutdown implemented. But my board running armbian seems to be negligent of this. On a side note, the CPU frequency reported in armbianmonitor -m is 1800/1300, in htop all 8 cores show 1300 ( .LITTLE)? expected behaviour: shouldn't the CPU be throttled automatically? excerpts from armbianmonitor -m: Stop monitoring using [ctrl]-[c] Time big.LITTLE load %cpu %sys %usr %nice %io %irq CPU C.St. 17:56:51: 1800/1300MHz 2.20 2% 0% 0% 1% 0% 0% 84.0°C 2/13 17:56:57: 1800/1300MHz 2.99 93% 0% 17% 75% 0% 0% 87.0°C 2/13 17:57:02: 1800/1300MHz 3.79 93% 0% 18% 74% 0% 0% 89.0°C 2/13 17:57:07: 1800/1300MHz 3.89 82% 0% 13% 68% 0% 0% 84.0°C 2/13 17:57:12: 1800/1300MHz 3.74 80% 0% 18% 60% 0% 0% 85.0°C 2/13 ... 17:59:19: 1800/1300MHz 9.83 90% 0% 16% 73% 0% 0% 98.0°C 2/13 17:59:25: 1800/1300MHz 10.38 95% 0% 15% 79% 0% 0% 99.0°C 2/13 Time big.LITTLE load %cpu %sys %usr %nice %io %irq CPU C.St. 17:59:30: 1800/1300MHz 10.27 94% 1% 22% 70% 0% 0% 100.0°C 2/13 17:59:36: 1800/1300MHz 9.85 90% 0% 16% 72% 0% 0% 99.0°C 2/13 17:59:42: 1800/1300MHz 10.10 95% 0% 17% 77% 0% 0% 100.0°C 2/13 17:59:48: 1800/1300MHz 10.33 98% 0% 18% 79% 0% 0% 102.0°C 2/13 17:59:54: 1800/1300MHz 10.38 95% 0% 22% 71% 0% 0% 101.0°C 2/13 18:00:00: 1800/1300MHz 10.79 97% 1% 20% 75% 0% 0% 100.0°C 2/13
  6. Figured it out, using pishrink, finally! It takes a while because 28 GB of empty space are being copied in the beginning, but anyhow ;-) This is my procedure: 1) boot from "plain" armbian (#2, see above) 2) insert original SD card (#1, 32 GB) in reader 3) dd if=/dev/sd#1 of=/path/to/image.img 4) /path/to/image.img 5)dd if=/path/to/image.img of=/dev/sd#3, #3 being the backup SD card, 16 GB note: "#1", e.g. is "/dev/sda", not "/dev/sda1", since you need to do a block level copy of the device, not only the disk.
  7. Update: I tried to clone the SD card doing the following, inspiration from this thread: 1) download armbian image, burn to another SD card (#2) 2) run armbian from this SD card #2 (works) 3) insert former boot SD card (#1) and backup SD card (#3) in SD card reader 4) dd if=/dev/#1 of=/dev/#3 bs=1M count=4 (to copy partition scheme and bootloader) 5) mount #1 /mnt/origin, mount #3 /mnt/clone 6) rsync -avSz /mnt/origin /mnt/clone (seems to work) 7) check /etc/fstab on #3, adjust UUID of / 8) try to boot from #3 --> doesn't work :-( I don't want to (can't) use dd because the SD card #3 is only 16 GB, whereas the #1 SD card is 32 GB, and my armbian installation takes only 4,1 GB. And I understand Igor's opinion that it's much more useful to save /home/* and /etc/* and a list of installed packages, but I want to have a working clone handy for instant use in case the #1 SD card fails, mostly because I am planning on installing pihole and I am worried about increased wear on the SD card. Can someone tell me why the approach aboce doesn't work? Does it have to do with /etc/fstab? Partitions? Thanks in advance! EDIT: both #1 and #3 are high quality (SanDisk Ultra A1)
  8. Hi! short question: Best way to create a bootable clone of sd card? I did some reading and found ">>create image on windows >> shrink image via gparted" doesn't work for me since I dont have a second computer running linux (will the linux subsystem on windows 10 help?). In the past, I used the rpi-clone on a raspberry, but reading in this forum has taught me that hotcloning an armbian sd card via script is not necessarily safe. Computers availabel to me are: 1) odroid hc2 running armbian 2) acer laptop running windows 10
  9. ...installed the JMS Firmware upgrade to be on the safe side and it seems I did not have the current version. See output of smartctl -a /dev/sdb1 below: === START OF INFORMATION SECTION === Vendor: JMicron Product: Generic Revision: 3101 Compliance: SPC-4 User Capacity: 4,000,787,030,016 bytes [4.00 TB] Logical block size: 512 bytes Physical block size: 4096 bytes LU is fully provisioned Logical Unit id: 0x3001234567891234 Serial number: 0123456789ABCDEF Device type: disk Local Time is: Tue Jan 29 11:45:38 2019 UTC SMART support is: Unavailable - device lacks SMART capability. Serial Number has changed, still no luck getting through to the HDD... EDIT: solved it! smartctl -a /dev/sdb1 gives the result pf the JMicron bridge, smartctl -a /dev/sdb1 -d sat gives the results of the HDD.
  10. Hi all! I am running a nextcloud server on an odroid HC2 and a 4 TB WD red HDD. So far, so good (thanks again @Igor for the quick help with the kernel downgrade!). And since I have bought the sbc recently, I do not seem to have the older firmware version with the HDD "emergency power down" bug, i.e. I haven't heard clacking sound from the HC2. Today, I set HDD spindown to 1 hour via hdparm (wd red does not support apm) because the HDD would spin down after about three minutes on its own. The HDD does not seem to apply these settings, so I installed smartmontools to check the HDD in general. Only to find that: smartctl -a /dev/sdb1 gives this: === START OF INFORMATION SECTION === Vendor: JMicron Product: Revision: 3101 Compliance: SPC-4 User Capacity: 4,000,787,030,016 bytes [4.00 TB] Logical block size: 512 bytes Physical block size: 4096 bytes LU is fully provisioned Logical Unit id: 0x3042000000000132 Serial number: DB00000000013B Device type: disk Local Time is: Tue Jan 29 11:03:47 2019 UTC SMART support is: Unavailable - device lacks SMART capability. To seems, it seems the JMicron USB-SATA Bridge prohibits armbian from seeing the actual HDD. Any way to fix this? Thanks!
  11. didn't change cables... downgrading the kernel did the trick! Thanks!
  12. oh, thanks! That was quick! which cables? the ethernet cable? So, it is rather unlikely the fstab entry with the 4tb hdd? I thought maybe the hdd is "too big" and checking the filesystem causes problems... EDIT: And how do I fix the network/downgrade the kernel? The migration is not finished yet, so downloading new image etc. wouldn't be a big problem. Should I do just that?
  13. Hi! recently bought a HC2 from hardkernel to migrate my nextcloud from a raspberry pi 3b. HDD is a WD red 4 TB. armbian stretch with the 4.19 kernel, downloaded from (Armbian_5.69_Odroidxu4_Debian_stretch_next_4.19.14). My problem: after boot (sudo reboot now or sudo shutdown -r), no possibility to login, i.e. angryip scanner doesn't show the hc2, putty times out when trying to connect. All LEDs are lit or pulsate respectively. I have to physically disconnect the power supply and reboot, then I can easily login. This is probably not very healthy for the hdd (or the sd card..., although with the ro option and zram, not much is being written to the boot section anyway, right?) I speculated that the hc2 has a problem with the 4tb hdd when booting and added the nofail option to /etc/fstab, but this doesn't do the trick. Any help is greatly appreciated. I really like the hc2 formfactor and after having used nextcloud on the raspberry pi for 12 months now, I'd like to upgrade to a more powerful board, but this problem seems like a dealbreaker if I can't fix it. EDIT: I corrected the used image, I actually had the 4.19 installed!