Corrupted GPT data drive partition. Please help with recovery.

1st Official Post

    Hello,


    I have been using OMV v3 on a Raspberry Pi 2 for almost 5 years now with no problems.

    I have 2 USB drive enclosures, each with 2 1TB laptop drives.... so 4 drives total with 4TB total storage space.

    The USB enclosures each have their own dedicated PSUs and the Pi has it's own dedicated PSU as well.

    All of the PSUs are plugged into a UPS.

    As far as I can remember, the drives were formatted as GPT EXT4 partitions, no LVM.

    This server has been up and running 24/7 for about 5 years, it's only been rebooted a few times for updates and to change the battery on the UPS once.

    It has never suffered an "unscheduled power down", all reboots and power downs were controlled and intentional.

    All it really does is store media files... music and movies that I play from a couple of other Pis running KODI.

    I have the maximum power-saving settings enabled on all of the drives, so none of the drives are spinning 24/7, they spin down and sleep after 10 minutes of non-use.

    I also have SMART monitoring enabled, and none of the drives have ever shown any warnings, they are listed as being healthy.

    Each drive has only one or 2 shared folders.


    This past week, I noticed I could no longer access 2 of the shared folders, these folders turned out to be on the same physical drive.

    I tried rebooting the server and drives, but the folders on that one drive were still not accessible from my clients, everything else seemed OK though.

    SMART monitoring did not indicate any problems with the drive.

    I decided to power down the server and pull the drive from the USB enclosure and check it on my Ubuntu desktop using a different USB drive docking station.

    The Ubuntu Disks utility sees the drive, but instead of showing the EXT4 partition and my files, the drive shows up as "Partition 1, 1.0 TB Unknown".


    I do not understand how this could have happened, especially when the SMART scan says the drive is OK.

    Unfortunately, this particular drive contains the backup of all my photography, business files, and my entire music collection, so I desperately need to recover the files if possible.


    I have tried various recommended analysis methods, fsck, gdisk, testdisk, etc.

    Testdisk recognizes the disk as an EFI GPT partition.

    However, I really don't want to proceed any further without some help and advice.

    Unfortunately I am also not able to copy the bad drive to another drive for testing purposes.

    Any attempt to copy the drive results in I/O errors from the bad drive.

    Yet when I run sudo smartctl -a /dev/sdc it shows no glaring problems with the drive.

    I desperately need to recover this drive intact if possible.
    Any advise would be most appreciated, preferably from someone who has experience and really knows what they are doing.

    I can't believe that the drive is damaged beyond recovery, it has just somehow lost it's partition information.

    Please help :(

    Green SMART is no guarantee that your drive is ok. There are lots of problems SMART doesn’t cover. Your drive has obviously died.


    I would simply replace it and recover data from latest backup.


    Also you should really consider OMV6. OMV3 is outdated since a very long time.

    I just upgraded to OMV5

    There is no OMV 6 install instructions or script for the Raspberry Pi apparently?

    If there is, can you please point me to a link?

    Will OMV6 even run on a 32bit R-Pi 2?

    I don't have any spare Pi's that are newer right now, and I can not afford to buy a Pi with the current price gouging that is going on, nor can I really afford to buy a 4-6 TB hard drive to do a full backup of the remaining drives on the server.

    Unfortunately I do not have a complete backup of everything that was on the failed drive.

    I probably only have about 70% of what was on there backed up, and the backups are on several DVD roms.

    I do not currently have a DVD drive, so I really need to get that data off that hard drive somehow.

    I know the drive is not "dead".

    It was working fine up until it wasn't, and I've never had a drive outright fail completely where most of the data couldn't be retrieved.

    It still spins up and can be read by testdisk and other diagnostic tools as I mentioned.

    The drive does not make any bad noises or clicking when it is powered on.

    It has simply lost it's primary partition record for some reason and I'm sure I could probably use one of the many backup partition records that testdisk says are still valid, but since I can not seem to copy the drive to another good drive, I am weary of writing anything to the bad drive.... hence why I asked if there was anyone with some experience in recovering data from a drive in this condition.

    So my original question still stands.
    Is there anyone who can please help or at least offer some instruction so I don't do any further damage to the drive?
    Thank you.

    I know it is a unfortunate situation, but what you say makes no sense at all. Every try to read from the drive results in I/o errors but you know it is not dead, because it was working fine before and makes no clicking noises? Bro, every drive was working fine before it died. Clicking noises may be an indicator the same way as bad SMART values are.


    What you can do is use dd with -noerror flag to force clone the whole drive to a new one while ignoring read errors. That is because every touching of the bad drive may result in even more data damage. You can now try to repair the broken filesystem and read data from the new drive without any risk.

    • Official Post
    Quote from HannesJo

    Green SMART is no guarantee that your drive is ok. There are lots of problems SMART doesn’t cover. Your drive has obviously died.


    I would simply replace it and recover data from latest backup.


    Also you should really consider OMV6. OMV3 is outdated since a very long time.

    I don't think 6 is available for a Pi 2.. I'm not even sure if 5 was.

    • Official Post
    Quote from Lucideye

    There is no OMV 6 install instructions or script for the Raspberry Pi apparently?

    If there is, can you please point me to a link?

    Will OMV6 even run on a 32bit R-Pi 2?

    Thread
    Installing OMV7 or OMV6 on Raspberry PI's, Armbian SBC's, & i386 32-bit platforms
    The following guides are for installing OMV7 and OMV6 on Armbian Buster, Raspberry Pi's and i386 platforms.

    **Notice to SBC users** While the repo's remain in place, OMV5 on SBC's is no longer supported.
    ___________________________________________________________


    Installing → OMV7 or OMV6 on Raspberry PI's
    (Applies to Raspberry PI's, models 2B and up.)

    Installing → OMV7 or → OMV6 on Armbian SBC's
    (Applies to most popular SBC models, the Odroids, Rock64Pro, Renegade, NanoPi, BananaPi and several…
    crashtest


    Yes, OMV 5 and 6 will run on an RPi2. It will be slow unless it is a late model RPi2 that actually has the same cpu as the RPi3.

    omv 7.7.9-1 sandworm | 64 bit | 6.11 proxmox kernel

    plugins :: omvextrasorg 7.0.2 | kvm 7.1.4 | compose 7.6.4 | cterm 7.8 | cputemp 7.0.2 | mergerfs 7.0.5 | scripts 7.2


    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!

    Quote from Jessy Berbers

    I would recommend buying an Odroid C4 as alternative.

    $190 for an SBC? Ludicrous to pay that much for something that costs pennies to make. Still too expensive for me right now. I am on a fixed income and have no spare money to spend at the moment. I need to work with what I have.
    Please stop with the useless comments and just answer my original question!!!!!

    Quote from HannesJo

    I know it is a unfortunate situation, but what you say makes no sense at all. Every try to read from the drive results in I/o errors but you know it is not dead, because it was working fine before and makes no clicking noises? Bro, every drive was working fine before it died. Clicking noises may be an indicator the same way as bad SMART values are.


    What you can do is use dd with -noerror flag to force clone the whole drive to a new one while ignoring read errors. That is because every touching of the bad drive may result in even more data damage. You can now try to repair the broken filesystem and read data from the new drive without any risk.

    I do not need your insults and attitude. If you can't offer help without insults and attitude then FUCK OFF
    I thought I made it clear that I don't have any other drives I can copy a full image to. I have one other available drive, and it is the same exact size as the drive that I am having trouble with, so none of the available rescue tools can copy an image because they tell me the target drive is too small. So If you really want to help, maybe you could send me a larger drive instead of your insults and attitude.

    Quote from Lucideye

    Please stop with the useless comments and just answer my original question!!!!!


    You asked for instructions on a repair attempt without risking further damage. That is not possible. How to do it with minimum risk of damage is the way I just described. A new drive is necessary for that. When you cannot effort it, you should put it aside and wait until you can. No offense, no insult, no attitude. Simple facts.

    • Official Post
    Quote from ryecoaaron
    Thread
    Installing OMV7 or OMV6 on Raspberry PI's, Armbian SBC's, & i386 32-bit platforms
    The following guides are for installing OMV7 and OMV6 on Armbian Buster, Raspberry Pi's and i386 platforms.

    **Notice to SBC users** While the repo's remain in place, OMV5 on SBC's is no longer supported.
    ___________________________________________________________


    Installing → OMV7 or OMV6 on Raspberry PI's
    (Applies to Raspberry PI's, models 2B and up.)

    Installing → OMV7 or → OMV6 on Armbian SBC's
    (Applies to most popular SBC models, the Odroids, Rock64Pro, Renegade, NanoPi, BananaPi and several…
    crashtest


    Yes, OMV 5 and 6 will run on an RPi2. It will be slow unless it is a late model RPi2 that actually has the same cpu as the RPi3.

    For some reason I remembered reading here (or somewhere) that Bullseye dropped support on some of the older devices, given the 2's age I assumed it was one of them. Apparently I was mistaken but as you said, it's going to be extremely slow.

    • Official Post
    Quote from Lucideye

    I do not need your insults and attitude. If you can't offer help without insults and attitude then FUCK OFF
    I thought I made it clear that I don't have any other drives I can copy a full image to. I have one other available drive, and it is the same exact size as the drive that I am having trouble with, so none of the available rescue tools can copy an image because they tell me the target drive is too small. So If you really want to help, maybe you could send me a larger drive instead of your insults and attitude.

    Damn man... I don't think you need PC help.. a therapist might be more in order. His post was not insulting at all. Not sure why you took it that personal.

    • Official Post
    Quote from Lucideye

    Any attempt to copy the drive results in I/O errors from the bad drive.

    What copy method are you using? Have you tried rsync by chance? A drive to drive rsync copy might be worth a try.
    _________________________________________________________________________________

    If you're getting I/O errors during a copy, the following note is true. There is no "perfectly safe" recovery.

    Quote from HannesJo

    You asked for instructions on a repair attempt without risking further damage. That is not possible.

    On the other hand, if nothing risked, nothing is gained.


    When it comes to "recovery":

    I've seen single drives lose significant numbers of files, or files were hopelessly corrupted during or after a recovery. Similarly RAID arrays have been "successfully recovered" with little to nothing in them. The only real insurance against data loss is backup. Generally speaking, the need for backup is a hard lesson users learn after they get burned.
    __________________________________________________________________________________

    1. Recover gpt partition table. (Assuming you can access the CLI.)


     sudo apt-get install testdisk


    Pick up the recovery on step 3 -> here.


    2. If you want to attempt a file system recovery -> fsck instructions.


    Note that there doesn't have to be anything wrong with the magnetic media of the drive. The SATA interface board of the drive can fail in any number of ways (near infinite) that SMART may or may not detect.

    In any case, good luck with the above.

    Quote from Lucideye

    $190 for an SBC?

    Take a look at the Odriod-HC4. It's under $100 and has everything needed for that price (a drive dock and a PS that will power up 5.25" drives). Disable the bootloader as noted -> here mid-way down the page. Add a hard drive for storage and an SD-card and you're in business.

    The OMV install process for the HC4 "ARMBIAN" is -> here .

    • Official Post
    Quote from crashtest

    Generally speaking, the need for backup is a hard lesson users learn after they get burned.

    This is so true. I was lucky and learned a long time ago... everything I lost, I was able to recover, just took forever.


    Vowed then it would not be a situation I'd ever be in again.

    • Official Post
    Quote from Lucideye

    I do not need your insults and attitude.

    The only attitude I read is yours. Keep it up and you will earn yourself a ban.

    omv 7.7.9-1 sandworm | 64 bit | 6.11 proxmox kernel

    plugins :: omvextrasorg 7.0.2 | kvm 7.1.4 | compose 7.6.4 | cterm 7.8 | cputemp 7.0.2 | mergerfs 7.0.5 | scripts 7.2


    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!

    Like 3

Participate now!

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

Register Yourself Login