Posts by tokul

    dmesg command in libreelec CLI will show, if sd device is recognized when you plug it in. It probably won't help, if your file system is setup on device and not on partition.

    Example of dmesg in spoiler

    Display Spoiler

    [348403.046290] usb 2-3: new SuperSpeed Gen 1 USB device number 3 using xhci_hcd

    [348403.065542] usb-storage 2-3:1.0: USB Mass Storage device detected

    [348403.066802] scsi host2: usb-storage 2-3:1.0

    [348404.073315] scsi 2:0:0:0: Direct-Access <<snip-hardware-name>> 1072 PQ: 0 ANSI: 6

    [348404.073705] scsi 2:0:0:1: Enclosure <<snip-hardware-name>> 1072 PQ: 0 ANSI: 6

    [348404.075072] sd 2:0:0:0: Attached scsi generic sg2 type 0

    [348404.075226] scsi 2:0:0:1: Attached scsi generic sg3 type 13

    [348404.079796] sd 2:0:0:0: [sdc] Spinning up disk...

    [348405.086129] ......ready

    [348410.154097] sd 2:0:0:0: [sdc] 1953458176 512-byte logical blocks: (1.00 TB/931 GiB)

    [348410.154530] sd 2:0:0:0: [sdc] Write Protect is off

    [348410.154533] sd 2:0:0:0: [sdc] Mode Sense: 53 00 10 08

    [348410.154785] sd 2:0:0:0: [sdc] No Caching mode page found

    [348410.154787] sd 2:0:0:0: [sdc] Assuming drive cache: write through

    [348410.338335] sdc: sdc1

    [348410.340926] sd 2:0:0:0: [sdc] Attached SCSI disk


    chkdsk in Windows CLI. Linux will not mount dirty NTFS.

    Did you turn off hybernation in that Seagate?

    I did a lot of web searching before I posted here. The vast majority do a bit of a pro & con list with zero evidence to support anything. Wikipedia have a very detailed entry if I was interested in technical specs. I'm not.

    If you are looking for cross platform compatibility, use NTFS. Both NTFS and exFAT are not native filesystems for Linux and NTFS implementation is older and more mature.

    Filesystem performance is less critical than device performance.

    Turn off hibernation in disk settings. If disks shutdowns on its own while being mounted, it can only cause mess. Disk should be stopped by OS and not other way. If your storage thinks that it is smarter than OS, you have to use synchronous file system without journal on it.

    > you probably supplied insufficient power to the drive.

    That's a distinct possibility. I've never seen so much damage done from that before.

    I had two 1TB WDs connected once. Was sorting files on them, drive started clicking, failed. Required CLI dskchk in windows to get it fixed.

    Technically they were connected to x86 pc and not to RPi, but they were sharing connection on same USB3 header and RPi is known to have limited power output on USB.

    Your story sounded like my two hour long dskchk multipled by six times.

    I won't try connecting HDD enclosures that rely on single USB connection to RPi or x86 USB sockets connected to same header.

    write your script without using that calculation.

    You can get current disk space usage in KB or MB. So *1024 part is redundant.

    Assumption that compressed package needs three times more to install is as inaccurate as hardcoding space requirement in a script. Try compressing bmp files and see how much they can expand.

    Nothing stops you from having several size checks and then setting required space accordingly.

    NVidia Quadro T400 ? Why put LibreElec on it?

    It might be economy version of Quadro card, but it is still from Quadro product line and new Nvidia cards tend to require binary/proprietary drivers from Nvidia.

    Which image is for USB stick boot? First one looks like disk failure or rpi is not giving enough power for argon one addon over USB. Second one is trying to mount device that does not exist in the system.

    some SD card writers come in "USB stick" form. If you could create image in USB stick, you can create image on SD card. Have you tried booting from SD card? Logical way for running LibreELec on RPi would be to install it on SD card and use SSD or NAS for storage.

    Sure, the NUCs are faster but as I stated in my initial post, top performance isn't really the goal. Also:

    * NUCs just don't hold up well to 24/7 usage. NUCs were my goto choice for media PCs but I have had two fail in the last couple of years and they are too expensive to be disposable. We also use them at work as SBCs at small remote sites and see high failure rates there too.

    * I am looking for a low power, passively cooled solution.

    there are many ways to interpret "don't hold up well". I was consider suggesting mini-itx based fanless mobo, but did not do that after checking NUC specs. You did sound like NUC was not powerful enough for you.

    I ended up getting a passively cooled mini PC based on the Celeron N5105 which is a quad core 10w CPU.

    Intel NUC 11 comes with Core i3-i7

    Celeron N5105 - 4 cores, 2.0 GHz, L2 cache1.5 MB, L3 cache 4 MB, UHD Graphics (24 EUs) 450-800 MHz, 2 × DDR4/LPDDR4X-293, 10 W

    Even I3 1115 got more cache and faster graphics support than that Celeron. i5 or i7 will outdo Celeron N5105 in cores.

    You just got yourself more options in case selection (if you went with mini-itx mobo and custom case), lower wattage and little less performance than NUC11.

    The only 5105 option that I see in local store is essentially non-Intel version of slower NUC at the Gen7 NUC price range.

    swap.conf is for making persistent configuration change which survives reboot.

    If you continue to use swap.conf, you would have to select swap file location which is accessible at the time when script is executed or change swap.service priorities. I am not the fan of debugging systemd and not the fan of using swap files, if it is not obvious at this point.

    You describe situation with USB media not mounted when swap file is created.

    Disk IO performance is about 100 times slower than memory performance. You would not want to have any memory read or write operations in file based swap.

    Please note that you are not running out of memory if free column in Linux free command has very small number. Your actual free memory is in last 'available' column. You will be running out of memory only when Linux starts killing running processes.

    Original poster was almost right about the way to activate swap manually. They only missed mkswap command and did not realize that swapon can activate individual swaps.

    1. dd to create file in writable location. Preferably not on SD card or SSD with limited write counters.

    dd if=/dev/zero of=/var/media/DISK_LABEL/swapfile bs=1M count=256

    2. format that file with mkswap

    mkswap /var/media/DISK_LABEL/swapfile

    3. swapon that file

    swapon /var/media/DISK_LABEL/swapfile

    those are main steps from LibreElec code responsible for stuff in swap.conf. Other steps only ensure that parent directory exists and secure swap file permissions.