Data Device error while searching for devices

  • Hi everybody,


    I'm new to this forum, and to OMV, too.


    I reactivated an old server, wanted to use it for all kind of SOHO stuff I currently have on an Raspi and a Buffalo NAS.


    For the NAS I was unsure to chose either FreeNAS or OMV, but decided for OMV as my skills with Debian are much better than with a BSD System.
    Base system is a Debian Stretch Xen 4.9.0 DomU on top of an AMD FX 8350, hypervisor enabled.


    For OMV I created a jessie Dom0, current state is OMV 3.0.88 Kernel 4.9.0-0.bpo.3-amd64.


    So far, so good.


    Now I'm unsure how to access my data drives (5x 1TB SATA).


    My idea is to create a RAID5 with one hotspare, LVM on top and hand that over to OMV.


    Maybe even without the LVM, any hints?


    However, for testing I tried to handover a raw disk, a mounted partition and even a LV.


    Guest sees the attached drive, but OMV not.


    When I try to detect the drives, OMV throws an error:


    Code
    Failed to execute command '[ -x /sys/class/scsi_host ] && for hostdir in $(find /sys/class/scsi_host -iname "host*" -type l); do echo "- - -" > ${hostdir}/scan; done' with exit code '1':


    /sys/class/scsi_host doesn't exist in my Dom0 which seems to be the reason for the exit code 1. After that OMV GUI is unresponsible, only a reload of the page help.


    I tried googling for the error, and searched here in the forum, but the results were all unanswered threads, so I'll give it another try for help :)


    Kind regards,
    Micha

    Backup: The duplicate copy of crucial data that no one bothered to make;

    used only in the abstract

  • Hi - I've got the same error and working on a resolution at the moment.


    Just figured out this is likely virtual machine related, I'm running Xen hypervisor with OMV on a virtual machine which can't see /sys/class/scsi_host at all. Running fdisk -l from OMV (shell) shows only the virtual disk so this isn't OMV related, I think its because the hypervisor (which can see the disks and /sys/class/scsi_host) isn't making the new disks available to the VM.


    Will post more when I figure out the solution!


    cheers
    Marcus

  • Got the first part solved by telling the host to make the physcial disks available to the guest. Now they show up in OMV


    virsh attach-disk <guest name> /dev/sda xvdb


    where /dev/sda is the physical disk on the host and xvdb is the virtual disk on the guest.

  • Well, I actually got an disk to show up in OMV, too, but can't create a FS on it.


    I don't use libvirt, but when I assign a LV to the DomU like this:



    Code
    [
                      'phy:/dev/vg_xen/omv-disk,xvda2,w',
                      'phy:/dev/vg_xen/omv-swap,xvda1,w',
                      'phy:/dev/vg_data/data,xvda3,w',
    ]

    it shows up as xvda3, too, like in your setup.


    There's still no /sys/class/scsi_host, but the error disappeared in OMV.


    Scan for physical disk is still negative.



    When I manually mount xvda3, I can see it under mounts in OMV, but on the wrong mount (it has to be named after the UUID as far as I found out).


    Currently I have no clue, I even was thinking about changing the setup (OMV as the main iron, other VMs under the vbox plugin in OMV, but that seems too dangerous, I doubt that OMV is yet mature enough to handle that, and when OMV breaks, all my systems are housed).


    I read too much about problems with upgrading OMV from 2.x to 3.x, and found myself so far a few quirks, like mandatory resolvconf package breaking resolv.conf, unnecessary ntfs-driver), I don't dare to put all my eggs in that basket.


    Still looking for a solution to run OMV in a Xen DomU, else I'll try something different.


    OK, before posting, I read the whole Thread again.


    You attached a new virtual disk, not just a partition like me.


    Thats good, now the disk shows up.


    I can even create a FS.


    So, all I need was to change xvda3 to xvdb. Actually very stupid from me to handle a disk over as a partition.


    So, ready for the next steps :)

    Backup: The duplicate copy of crucial data that no one bothered to make;

    used only in the abstract

  • Hi,


    Am glad at the end you got it working. Just to report i'm using a clean install of OMV on a Xen DomU. Had to do a net install of Debian and then install OMV "on top" as there is no net install of OMV but it all worked OK (instructions online). OMV was first installed as 1.x and then I did a double upgrade from a root shell to 3.x before configuring anything.


    When I got my disks visible to the DomU I was able to see them in OMV, create a RAID 1 setup and then put a btrfs filesystem on top. All works well through the admin webpage. Have been able to setup users and shares but left with 2 issues:


    a) booting the OMV DomU just takes me to a grub prompt. Not big issue as I can type in set root=(hd0,1) etc to get it to boot OK
    b) can't see my OMV system from other computers on the same network. If I address it directly e.g. \\192.168.x.x then I just get asked for user & password even for a public share. Will figure that one out as soon as I get more time & less sleep......


    cheers
    Marcus

  • Am glad at the end you got it working.

    At least I'm a step further. Disks and FS seems to be OK
    However, shares are not yet working as expected, neither with SMB nor with AFS.
    I'll fiddle around with it …



    a) booting the OMV DomU just takes me to a grub prompt. Not big issue as I can type in set root=(hd0,1) etc to get it to boot OK

    Are you using pygrub?


    Default setup from Stretch, no issues at all.




    When I got my disks visible to the DomU I was able to see them in OMV, create a RAID 1 setup and then put a btrfs filesystem on top.

    I'm also interested, are you not using LVM in the DomU?

    Backup: The duplicate copy of crucial data that no one bothered to make;

    used only in the abstract

Jetzt mitmachen!

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