Problems with Corrupted data

  • Hi all, I have a configuration with 4x4TB WD disks in a OMV 5.5.11.1 installation, upgraded in time. I configured a RAID5 with an 11 TB ext4 file system, and actually there are 8 TB of data. The OMV installation is made under PROXMOX with the configuration as for image, and I attached the disks via Marvell PCI-E 4 sata card


    I noticed something that scaries me, this is the scenario:


    I have some splitted archive with winrar in another disk, I test It and everything is ok. When I copy this archive on the ext4 filesystem, I try to verify or extract the archive and I receive EVER a data corruption in some of these packages. I copy data usualyy with windows explorer.


    So, absolutely there's a problem somewhere I cannot understand, and I think that for sure some of my old data inside this pool is corrupted.


    I attch some screenshot abou my installation.


    If someone could help me to try to investigate the issue.


    Thank you

  • Yes I did, I don't have any error...now I'm going to investigate on SMB because I did a configuration to preserver old devices compatibility like this:


    client min protocol = NT1

    ntlm auth = ntlmv1-permitted


    I'm going to copy data for example via FTP and test

  • I did a configuration to preserver old devices compatibility like this:


    client min protocol = NT1

    ntlm auth = ntlmv1-permitted

    I don't think the above has anything to do with corruption. NT1 is an older authentication protocol.


    Does the OEM have a utility for checking / testing the Marvel PCI card?
    (I'm assuming you got the card for 4 more SATA ports and that you're not using server grade hardware.)

  • I don't think the above has anything to do with corruption. NT1 is an older authentication protocol.


    Does the OEM have a utility for checking / testing the Marvel PCI card?
    (I'm assuming you got the card for 4 more SATA ports and that you're not using server grade hardware.)

    I don't know, yes It's a simple PCI card with sata ports, the brand is SYBA

  • I tried everything...Using winrar or 7zip---splitted archive or single...Changing the virtual eth on Proxmox...Everytime I copy my archive on OMV ext4 fs, I have a CRC error...really frustrating...

  • The problem is, with a RAID configuration, the SYBA / Marvel card has become a single point of failure - the 4 ports can't be bypassed. Without an OEM utility to test the Marvel card, it would be nearly impossible to eliminated it as a suspect. Further, I'm assuming that you don't have 100% backup or you wouldn't be posting. (If that's the case, bad news.)


    What you could do (this is spit balling):


    - Run a long SMART tests on all drives. One of them may have a developing problem that you're unaware of.
    - If you captured a past version of the OMV VM, in Proxmox, try restoring it. (OR)
    - Rebuild OMV VM from scratch and reconnect to your existing data, recreate network shares and test.
    - Add an extra drive to the host, format it etc., and do a local copy from the command line, using something like MC. (apt-get install mc) That would narrow it down to platform hardware, eliminating any network possibility.
    - If you configured software mdadm RAID, there's a chance that something like a live distro (for example Knoppix) booted from a CD or USB thumbdrive might recognize the array. It seems as if you have enough Linux knowledge to configure up a quick network share, on a live distro, to see what happens in a copy operation. If that works, it might be time to look at Proxmox.


    The above is free, but time intensive. Beyond the above and in the bottom line, this could be the MOBO, the power supply, the marvel card or even one of the hard drives. There's no knowing for sure without testing and, given that you've combined drives into RAID5, testing individual drives would be destructive.

    This "seems" like a hardware issue. If I were to speculate, and not this is Speculation (with a capitol S), it might be the SYBA card. Without care during installation, using a grounded cuff or insuring the skin contact is made with a grounded chassis at all times, ESD damage is possible and it could manifest like this (intermittent.)

  • Hi crashtest, I think It's a fantastic answer!!! Thank you!


    I'm goint to start one by one what you suggested, I have time and I want to investigate more this issue. Going to launch long self-test on each drive, I'll keep you updated.


    Thank you :)

  • When I saw your initial post I knew you would suggest a plan of action, my thought on this is from post 7, CRC errors, these are usually associated with networking,

    If this is the case (I've never experienced it personally), vatastala it might be better to start with a locally installed hard drive or a USB external, do come copy operations to eliminate or confirm that it's the network.

    If it's attributable to the network, don't forget to swap out cables and check connections. (The latter, a simple connection, got me just recently.)

  • If this is the case (I've never experienced it personally), vatastala it might be better to start with a locally installed hard drive or a USB external, do come copy operations to eliminate or confirm that it's the network.

    If it's attributable to the network, don't forget to swap out cables and check connections. (The latter, a simple connection, got me just recently.)

    Hi,


    after the long SMART test, I confirm that there are no problems on the disks, seems that everything is ok.


    Regarding Network, I have a SINGLE cable from my router to the Host machine, so all the rest of the network is made by Virtual Network devices in ProxMox.


    So, I did other tests as for the following scenario:


    - On ProxMox I have 2 VMs, OMV and a Windows 10

    - On OMV Vm I have the 8GB OS disk and the 12TB drive attached by Syba SATA card.

    - We know that everytime I copy an archive from Win10 to the SMB share on the ext4 SMB partition, I have a CRC error, so I createad e new one, copied, tested....CRC Error

    - I copied via WinSCP the archive on OMV OS disk, in home dir from my Win10 Vm, so via Network

    - I downloaded It again via WinSCP on the Win10 machine, tested and extracted...all ok

    - Connected via SSH to the OMV Vm, I copied the archive from the OS disk to my mounted ext4 volume, tested from the Win10 machine...CRC ERROR!!!

    - Last test....copied the archive via WinSCP from the Win10 machine to the mounted ext4 partition inside OMV Vm, tested and extracted the archive from Win10 machine with Winrar...THE ARCHIVE IS OK!!! So, with WinSCP every copy operation is a success, with SMB no.


    So what now? I think I don't have to change the unique Eth cable, infact for every operation outside this It's ok, the problems maybe are the SATA cable on the Syba card? How could the Syba card be the problem if I don't see particular errors in the OMV virtual machine console and via WinSCP the copies are ok?


    EDIT: mmm no, after other tests I have archives damaged also after WinSCP copies...So now I have to do some tests copying files from USB, If ok maybe install a new OMV VM and attach Syba card OR try Knoppix live distro...what do you think?

  • Well, it's not a network error unless something is going on with the Proxmox virtual switch (simulated networking in software). Along those lines, Proxmox has a user forum where you might search on CRC errors. In any case, whatever is going on is internal to the host and it's VM's whether it's the Syba card, other hardware, or the various software builds. While this may not help, I've never heard of SAMBA generating CRC errors.


    I'm still speculating that this is hardware, not software, but the only easy (and free) options are complete rebuilds from scratch.


    I'm guessing you have a reason for setting this up, in this way, but there are additional layers of complexity involved in virtualization. Further, backing up a complete "Proxmox platform" requires resources a typical home user doesn't have..

    It might be best to get a Win10 client and run OMV on bare metal (without the Proxmox hypervisor). Along those lines, there are plenty of older servers on the used market for running OMV, but it takes a lot of research.


    What model of Syba card do you have? Have you been to the OEM's support web site?

  • This is the card:


    https://www.sybausa.com/index.…=64_181_85&product_id=156


    I understand what you say, but for the moment I'd like to investigate further trying your other suggestion, I'm going to download knoppix live and see If It sees the configuration

  • I'm going to take it that you did the manual Linux drive install, for the board, in "Proxmox". (Since Proxmox is presenting drives to OMV.)

    Yes sure, I configured Proxmox to present this PCI card to each VM I want to configure, simply adding a line in the VM configuration with the PCI address of the card itself

  • I installed OMV last version from scratch and the problem seems gone, I'm doing many many tests but no more errors...the only differences between the old OMV Vm and the new are:


    - The old VM was upgraded different times

    - The VM conf is a little different on ProxMox, the main differences are the kind of VM Machine ( It's a ProxMox parameter ) and the bios ( old VM UEFI, the new one DEFAULT SEABIOS ).


    For the rest, everything is the same...So, I really really don't understand what happened during time, and how the OMV installation, even If seems perfectly working, can cause these kind of issues.

Participate now!

Don’t have an account yet? Register yourself now and be a part of our community!