Unable to restore 4k cluster size NVME



  • So fog has been working great until I tried to deploy a win 10 image captured from my NVME. The NVME was converted to 4k cluster size from back in the day when Sierra supported no other on NVME.

    This is what happens when I try to deploy the captured image onto a UEFI based VM.

    alt text

    Then this

    alt text

    I am thinking about converting the drive back to 512 k cluster…



  • @Sebastian-Roth

    Also I just updated to 1.58 it went smoothly. I was worried when it forced a password on mysql, I screen capped it and so far everything is working.

    This kind of thing is so much more fun compared to Windows stuff.



  • @Sebastian-Roth

    Oh. Got it. That makes sense.


  • Senior Developer

    @Fog_Newb said in Unable to restore 4k cluster size NVME:

    The one thing I was curious about is why a 250GB drive would be needed since the image is around 60GB

    Well I did expect this to be a non-resizable image. If you have a resizable you’d need to also convert d1.minimum.partitions file. I’d think that should work just as well.



  • @Sebastian-Roth

    This is very cool. The image finished deploying and it can boots up Windows but it would not load the desktop it goes to repair. But this could be because there is a huge difference in hardware?

    Perhaps a feature could be added to convert back and forth from 4k to 512 in the web interface. The one thing I was curious about is why a 250GB drive would be needed since the image is around 60GB


  • Senior Developer

    @Fog_Newb Good to hear! You can do it the other way round as well. Capture from a 512 sector size disk, divide all the numbers by 8 and deploy to a 4k disk.

    It shoudn’t be too hard to implement that for FOG 1.6 I reckon.



  • @Sebastian-Roth said in Unable to restore 4k cluster size NVME:
    Cool! I changed d1.partitions to what you posted and it was able to deploy the image onto a VM with a 512 sector drive.

    alt text


  • Senior Developer

    @Fog_Newb said in Unable to restore 4k cluster size NVME:

    This image was captured from the 4k block NVME

    Yeah right, I see. Converting this to 512 byte block size is as easy as multiplying every sector count number by 8 - result:

    label: gpt
    label-id: 479D53EA-F9B2-4E50-97A5-3C89ACD97D43
    device: /dev/nvme0n1
    unit: sectors
    first-lba: 48
    last-lba: 500118144
    
    /dev/nvme0n1p1 : start=        2048, size=     1083392, type=DE94BBA4-06D1-4D40-A16A-BFD50179D6AC, uuid=7ED58F36-04DC-4D18-9CD2-C7D832B9F61E, name="Basic data partition", attrs="RequiredPartition GUID:63"
    /dev/nvme0n1p2 : start=     1085440, size=      614400, type=C12A7328-F81F-11D2-BA4B-00A0C93EC93B, uuid=A6A8E7DC-0655-4E99-A1FB-5F42A76303A7, name="EFI system partition", attrs="GUID:63"
    /dev/nvme0n1p3 : start=     1699840, size=       32768, type=E3C9E316-0B5C-4DB8-817D-F92DF00215AE, uuid=2BBCB147-BA4C-4545-806A-57770D209F01, name="Microsoft reserved partition", attrs="GUID:63"
    /dev/nvme0n1p4 : start=     1732608, size=   498384896, type=EBD0A0A2-B9E5-4433-87C0-68B6B72699C7, uuid=BECEA3AA-9442-42BF-B497-A3E0FF5D7D94, name="Basic data partition"
    

    Should fit on a 512 byte sector size disk with 250 GB roughly (a little bit less would even do).



  • d1.partitions.txt @Sebastian-Roth said in Unable to restore 4k cluster size NVME:

    I have since delete the BIGPIG image. Here is the one from TEST. This image was captured from the 4k block NVME

    d1.partitions.txt


  • Senior Developer

    @Fog_Newb Notice the size difference? 555 divided by 70 equals roughly 8, same as 4096 / 512. There is no point in adding the -C option because the partition layout is not made to be pushed out to a 4k block size disk.

    Can you please post your d1.partitions file you find in /images/BIGPIG/ or /images/Test/ on your FOG server? Possibly we can convert the numbers manually and try deploying like that.

    I found scripts that I used some time ago to get 4k virtual disks. Turns out this was not on creation time of the VM disk but when starting the VM itself. I used QEMU with parameters -global ide-drive.physical_block_size=4096 and -drive file=disk.qcow2,if=ide,index=0,media=disk,io=native,physical_block_size=4096. Though I have not played with this in a long time and can’t remember exactly but I think it worked.



  • @Sebastian-Roth

    Thanks. Do you know where I can add the -C option to skip the disk size check?

    alt text


  • Senior Developer

    @Fog_Newb Back then I used qcow VM disk Image format I think. Not sure though. I will see if I can find some old scripts or documentation.



  • @Sebastian-Roth

    I have been searching for information on how to create a 4k sector virtual drive in VMWARE workstation and ESXi 6.7 but I can’t find it. Any ideas?


  • Senior Developer

    @Fog_Newb This is a good topic. We have looked into that some time ago but as still pretty much all disks are 512 cluster size we never got to the point of making FOS capable to do this. I remember trying to setup a virtual VM disk to use 4096 cluster size to be able to able to even work on this.

    Let’s keep this topic open and see if other people come around want to see that new feature as well.

    Moving to feature requests section.



  • @Sebastian-Roth

    It looks like what went wrong is… When i tried to deploy an image that was captured from an NVME with a 4k block / sector size on a pc with a 512 block /sector size it gets confused about the size of the image. Like in the screen shot, it shows the image is 300+GB but it was really 30 something GB.

    Also, trying to deploy an image captured from a 512 sector sized drive to the 4k sector NVME … it will deploy onto the drive but windows won’t boot on it.

    So I just figured you can’t image between drives with different sector sizes.

    I had converted the drive to 4k back in the day so Sierra could run native on it.

    alt text

    I now want to go back to 512 but unfortunately I didn’t think it through prior to reloading the OS from scratch last night.

    alt text


  • Senior Developer

    @Fog_Newb Why delete? It’s better you share what went wrong and how you solved it as this might help other people in the future! :-)



  • uh… never mind please delete


Log in to reply
 

292
Online

7.1k
Users

14.3k
Topics

135.0k
Posts