Problem with system rescue (systemrescuecd) when you have separate /boot partition

  • Like many I started with a separate, small /boot partition.

    But, then I discovered that OMV has the System->Kernel menu that permits one to install system rescue cd (and others).

    So, first I had to increase my /boot partition size. A bit fiddly because my root,var,home,swap were all in LVM. But, done.

    Then I activated the insta;;/download and the BOOT ONCE option.

    Rebooted. Menu came up and stayed, highlighting the rescuecd option.
    I wanted it to timeout like my normal boot. That required setting

    GRUB_RECORDFAIL_TIMEOUT=$GRUB_TIMEOUT

    in /etc/default/grub.

    Good.

    Next, it DID timeout but failed because of some gobledygook that grub complained of.

    Took me a while.

    There is a file in /etc/grub.d/

    42_sysresccd

    In there I looked. First there is a "probe" command - for which I could find no documentation for grub2.
    The comment I got from chatgpt was "

    In GRUB 2, the equivalent of the probe command for setting the root device UUID is typically handled automatically by GRUB's built-in capabilities. You don't usually need to manually probe for the root device UUID as it's detected during boot.

    "
    Which implies the command is no longer valid and that something else should be used instead. However, it DOES WORK. Maybe it will disappear from grub2 silently? Food for thought.

    So, why wasn't it booting?

    Because the variable isofile was set to /boot/system......blahblah - the iso file name.
    AND - because the /boot is mounted/seen already - one should not put /boot - only /system....blahblah.
    The iso file path is RELATIVE to the mount point of /boot.

    Last - I wanted to install nomedia server. That would require (a) locating the package for arch linux (the os that systemrescucd uses) and (b) having some permanent (backing) store so that it would be there with all other programs/customisations I might make.
    (b) was easy - create a LABEL for the partition that /boot is on. My case I used tune2fs -L OMVBOOT.
    and then edit the 42_sysresccd file in /etc/grub.d and add in cow_label=OMVBOOT as one of the kernel options. I placed it after the "copytoram".
    That then creates an overlay with all my mods in omv /boot/persistent folder. (It creates the folder name - you do nothing). So, one can even revue one's changes when OMV is up and running/.
    (a) Was more difficult as I am a complete Arch novice.

    Normally top install programs/packages, Arch uses a command (as root of course) called pacman.
    To install an Arch supported package from their repositories - you go 'pacman -S packagename'.

    BUT nomachine isnt there.

    They have a user supported repository that uses another method. It's called AUR. So, any package that has "AUR" in it - can't be straight installed by pacman.

    What you have to do is download it from git and then "makepkg". Which builds/compiles it from source and you end up with a tar file ending in .tgz.gst.
    You can then run pacman -U thing.tgz.gst and it will install.

    I won't bore you with the details of setting up an Arch VM in virtulbox (really, really bad install system IMHO).
    But, in the end, I got the nomachine.tgz.gst, copied over to my OMV system /boot and then booted to systemrescue (via System-.Kernel)

    and ran pacman -U.

    Tweaked nomachine network to start automatically - for some reason on Arch it didn't set it up with it turned on. Minimalist Arch attitude - I guess.
    useradd - a user for myself with password. Not that I will use it - but, using nomachine client - one has to have a user that's not root in order to connect and then view/play with the mirror of the desktop.


    One last thing - as part of nomachine install it implied it could work without X/wayland being setup. Which, in turn, means I wouldn't need a graphic card.
    Some graphic cards with HDMI won't boot up properly if a screen isn't attached. I know one can get an HDMI dongle to solve that. But, my graphics card is very old and displays in yellow and is slowing down my pcie bus speed AND occupying a 16 lane slot which I could use with my NVME 4 x M.2 pcie card.
    I guess (hopefully) that the nomachine package knows how to do xvfmb (virtual frame buffe/pretend graphics card) and so, I will see whether or not it will all run TRULY HEADLESS.

    I am a pensioner and even a cheap 2ndhand graphics card is expensive to me.


    Sorry to woffle on. Hopefully - some of this maybe helpful to someone.

    :)

  • chente

    Hat das Thema freigeschaltet.
    • Neu
    • Offizieller Beitrag

    Like many I started with a separate, small /boot partition.

    But, then I discovered that OMV has the System->Kernel menu that permits one to install system rescue cd (and others).

    So, first I had to increase my /boot partition size. A bit fiddly because my root,var,home,swap were all in LVM. But, done.

    Having a separate boot is not normal for OMV when installing with the OMV iso. There are an infinite number of ways to install debian and I chose to support the normal OMV install. I will look to see if I can make the iso storage location a variable or something but the plugin has been doing this since the beginning (even back when it was part of omv-extras).


    In there I looked. First there is a "probe" command - for which I could find no documentation for grub2.
    The comment I got from chatgpt was "

    In GRUB 2, the equivalent of the probe command for setting the root device UUID is typically handled automatically by GRUB's built-in capabilities. You don't usually need to manually probe for the root device UUID as it's detected during boot.

    "
    Which implies the command is no longer valid and that something else should be used instead. However, it DOES WORK. Maybe it will disappear from grub2 silently? Food for thought.

    The plugin has been around a long time. I test each new version of OMV when I port the plugin. If it works, I don't change anything. The parameter may disappear but not until the next version of Debian since Debian is a stable release. So, I will catch it when it disappears. That said, I will look at updating the config.

    Because the variable isofile was set to /boot/system......blahblah - the iso file name.
    AND - because the /boot is mounted/seen already - one should not put /boot - only /system....blahblah.
    The iso file path is RELATIVE to the mount point of /boot.

    This is only accurate for your system. Setting a standard omv install to /systemrescue.iso will not boot. Even the systemresuce docs use the full path - https://www.system-rescue.org/…SystemRescue_on_the_disk/

    omv 7.0.5-1 sandworm | 64 bit | 6.8 proxmox kernel

    plugins :: omvextrasorg 7.0 | kvm 7.0.13 | compose 7.2 | k8s 7.1.0-3 | cputemp 7.0.1 | mergerfs 7.0.4 | scripts 7.0.1


    omv-extras.org plugins source code and issue tracker - github - changelogs


    Please try ctrl-shift-R and read this before posting a question.

    Please put your OMV system details in your signature.
    Please don't PM for support... Too many PMs!

Jetzt mitmachen!

Sie haben noch kein Benutzerkonto auf unserer Seite? Registrieren Sie sich kostenlos und nehmen Sie an unserer Community teil!