OVM doesnt start anymore

1. offizieller Beitrag

    Hello, first of all, I do not know Linux so well, I'm just learning. Now I have the following problem, after a mainboard change I get this error message:


    mdadm found some drive for an array that is already active: /dev/md0 Mdadm giving up

    Gave Up waiting for root filge system device. Common Problems:

    -Boot args (cat /proc/cmdline)

    - Check rootdelay = (did the system wait enough?)

    -Missing modules (cat /prox/modules ; is /dev

    ALERT! /dec/sda1 does not exist . Dropping to shell


    I have connected all hard drives and changed all cables no result. Maybe one of you can help me ?


    Greetings

    Trinan


    Translated with http://www.DeepL.com/Translator (free version)

  • macom

    Hat das Thema freigeschaltet.

    Plug sda to another port on the mainboard and try again.

    If you got help in the forum and want to give something back to the project click here (omv) or here (scroll down) (plugins) and write up your solution for others.

    Since you changed the Mainboard, you can try to only plug the OMV drive and boot up.

    You probably will need to update-grub


    See if all is OK apart the missing drive's array.

    After, you can plug the rest of the drive's and boot up again.


    Now, some speculation:

    mdadm arrays are bound to the hardware they are built, (AFAIK).

    Changing from one Mainboard to another implies a change of controller unless the Mainboards are exactly the same.


    This is something that geaves can shed some light on (I hope)

    Another stupid question: new motherboard, but OMV4. Is the board supported by debian stretch?

    If you got help in the forum and want to give something back to the project click here (omv) or here (scroll down) (plugins) and write up your solution for others.

    • Offizieller Beitrag
    Zitat von Soma

    mdadm arrays are bound to the hardware they are built, (AFAIK).

    Changing from one Mainboard to another implies a change of controller unless the Mainboards are exactly the same.

    This shouldn't be an issue, I switched my Intel server for the HP Microserver, the only config change was the networking and for that I ran omv-firstaid direct on the console, made sure everything was working, reboot to test the boot drive, shutdown and connect data drives, at that time I was using Raid 5.

    Zitat von geaves

    This shouldn't be an issue, I switched my Intel server for the HP Microserver, the only config change was the networking and for that I ran omv-firstaid direct on the console, made sure everything was working, reboot to test the boot drive, shutdown and connect data drives, at that time I was using Raid 5.

    I really thought that Hardware RAID was binded to the controller it's built on.

    That was the reason what made me make a Software BTRFS RAID on my IcyBox (used the JBOD mode of it).


    If the Hardware fails, I can simply put the 2x drives on another box/USB to SATA adapter and the system will be seen as normal.


    Well, we're always learning, :)


    So I only connected the hard drive with the OMV, after that went. After connecting the other hard drives, everything worked as usual. Now I just have to find the OMV in the network, thanks for the help <3

    • Offizieller Beitrag
    Zitat von Soma

    I really thought that Hardware RAID was binded to the controller it's built on

    It is and it isn't, a hardware raid controller such as these look after and control the drives attached to it, these have their own bios and load during post.

    M'boards that have X number of sata ports that have the capability of running in a raid configuration, (there's a name they give to this) and are set up in the m'boards bios. If you use a m'board raid the drives are presented to likes of OMV as a single device/drive, and yes you have to use an identical board.

    To use mdadm (software raid) you need to disable the raid option on the m'board so that the drives are presented as single drives to the OS. The real bummer comes with the likes of my N54L which uses a backplane, this is like a sata hat board, but it then has single cable connection to the m'board, this forces the user to run raid as there is no way to disable it in the bios. What I had to do was to install an 'adjusted' bios :) which displays all the hidden configuration and disable the raid option.


    Qnap and Synology use BTRFS on mdadm and in some of the earlier models it would create an LVM.


    I've helped a couple of users on here that have used BTRFS and XFS on mdadm, some of the cli commands are totally different to ext4

    • Offizieller Beitrag

    I have always seen this topic in a very simple way.


    If the disks are controlled by hardware that creates a Raid (whatever), the operating system will not see the disks, it will see the Raid. To recreate it you will need the same driver / hardware.


    If the hardware lets the operating system see the disks independently, the operating system can create a Raid (of whatever type) with those disks, and this can be recreated on another system with the same operating system.


    In the end it will be irrelevant if the disks are connected by SATA, by USB, inside a box or wherever. The question is whether that hardware allows the operating system to view the disks independently or not.


    I came to this conclusion many years ago. geaves correct me if I'm wrong please. Sometimes you make me doubt basic things. :)

    • Offizieller Beitrag
    Zitat von chente

    If the disks are controlled by hardware that creates a Raid (whatever), the operating system will not see the disks, it will see the Raid. To recreate it you will need the same driver / hardware

    Yes, the drives are presented to the OS as a single drive. The downside to this with Linux particularly headless servers unless the hardware raid bios has some option to send emails, you have no idea the state of the array not the drives within. Hardware Raid controllers are great in a Windows environment as most boards have their own software which 'reads' the bios on the card and allows a sys admin to monitor it by email alerts, but even in Windows the array is still presented as a single drive.

    Zitat von chente

    If the hardware lets the operating system see the disks independently, the operating system can create a Raid (of whatever type) with those disks, and this can be recreated on another system with the same operating system.

    Yes, the advantage with software raid is the array can be 'moved' to another system, as in the OP's case, because mdadm has written a signature to each drive, can it moved to another OS? that's something I've not tried.


    What I would find interesting is the likes of Soma IcyBox, if this was set as a raid the two drives would present themselves as a single drive to the OS, however how do you know the state of that array, unless they are equipped with an IPMI usually through an Ethernet port. If they only have USB there may be another option to monitor the array, but these things are designed and built for Windows, even though they say Linux compatible (compatible being rather broad)

    • Offizieller Beitrag

    Thanks for confirming, I was already beginning to doubt my beliefs. :)

    Zitat von geaves

    What I would find interesting is the likes of Soma IcyBox, if this was set as a raid the two drives would present themselves as a single drive to the OS, however how do you know the state of that array, unless they are equipped with an IPMI usually through an Ethernet port. If they only have USB there may be another option to monitor the array, but these things are designed and built for Windows, even though they say Linux compatible (compatible being rather broad)

    This case for me boils down to the same principle. If the firmware in the IcyBox box "allows" the operating system to view the discs independently, the operating system will be able to create a playable Raid with that same operating system. If it is the firmware of the box that creates the Raid, this same firmware will be needed in another system to handle this Raid. Another issue is, as you say, that the box creates the Raid and also allows the operating system to obtain details of that Raid. But this is independent.

    Some RAID controllers allow to define JBOD / RAID 0 with only one disk, even multiple of these. This will allow the OS to see single drives and build an array out of it. But most of the time these "drives" will not support SMART.


    Had that some time with a Microserver and the internal B120 RAID controller. (But it can be switched to ahci mode which allows to see single drives in the natural way).

    If you got help in the forum and want to give something back to the project click here (omv) or here (scroll down) (plugins) and write up your solution for others.

    Zitat von geaves

    What I would find interesting is the likes of Soma IcyBox, if this was set as a raid the two drives would present themselves as a single drive to the OS,

    This case in particular has 4 modes of configuration via DIP switches: JBOD, RAID0, RAID1 or BIG.

    Since I assembled the drives as JBOD, I see them as 2 individual drives either on Windows or on Linux.

    I have no idea how they would show if a RAID mode was selected (and to test it, I would need 2 new drives. The Box cleans any drive that is plugged inside if the DIP switches are changed)


    The chip that is on the case is the ASM1352R (where R stands for RAID). Altough very reliable and fast, it's a hardware RAID binded to that specific equipment.

    How would the OS deal with the RAID? That is something for Jeff Gerling to test, maybe, :)


    chente

    That was the reason I went with software RAID, so I wouldn't be dependent of the same case/controller in case of failure.

    Software errors, I can solve (rather) easely but, with hardware, I'm always supicious.


    Zoki

    This one only see's the physical drives (AFAIK)


    If you guys want to see it better (the datasheet is on the website), here is the link to the box:

    ICY BOX - Always well connected.



    On a side note, I got a few weeks ago another model, IB-3740-C31 but this one is only 4xHDD on a JBOD.

    The chip used is a Elan EM88F758N which is handling pretty well 3xHDDs and 1xSSD.


    Well, I'm not that demanding on disk I/O so, take this with a pinch of salt.

    The USB box (although recognized as UAS) only goes till 5Gps.

    • Offizieller Beitrag

    Yeah, the supported OS's being Windows and Mac, that would suggest there is some sort of software available to monitor a raid setup.


    But this made me smile -> RAID 5 therefore offers increased fail-safety.

    now I have the problem that my OMV is no longer recognized in the network. can also not access the Webui :( you should do what with SSH but have no idea how that works

    Zitat von geaves

    IcyBox RAID: how do you know the state of that array

    If a drive is faulty the related drive LED would go red instead of the normal blue. There is no other indicator, but non-technical users are happy with this and can fix such an error easily.

    omv 6.9.6-2 (Shaitan) on RPi CM4/4GB with 64bit Kernel 6.1.21-v8+

    2x 6TB 3.5'' HDDs (CMR) formatted with ext4 via 2port PCIe SATA card with ASM1061R chipset providing hardware supported RAID1


    omv 6.9.3-1 (Shaitan) on RPi4/4GB with 32bit Kernel 5.10.63 and WittyPi 3 V2 RTC HAT

    2x 3TB 3.5'' HDDs (CMR) formatted with ext4 in Icy Box IB-RD3662-C31 / hardware supported RAID1

    For Read/Write performance of SMB shares hosted on this hardware see forum here

    Zitat von Soma

    That was the reason I went with software RAID, so I wouldn't be dependent of the same case/controller in case of failure.

    what I found is that other cases with ASM chipsets (likely within the same family of chipsets) can read & write. Had posted this in the past here

    omv 6.9.6-2 (Shaitan) on RPi CM4/4GB with 64bit Kernel 6.1.21-v8+

    2x 6TB 3.5'' HDDs (CMR) formatted with ext4 via 2port PCIe SATA card with ASM1061R chipset providing hardware supported RAID1


    omv 6.9.3-1 (Shaitan) on RPi4/4GB with 32bit Kernel 5.10.63 and WittyPi 3 V2 RTC HAT

    2x 3TB 3.5'' HDDs (CMR) formatted with ext4 in Icy Box IB-RD3662-C31 / hardware supported RAID1

    For Read/Write performance of SMB shares hosted on this hardware see forum here

    Danke 1

Jetzt mitmachen!

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

Benutzerkonto erstellen Anmelden