Please point me to 1.2.0 -> 1.3.4 for Centos 6.7 upgrade docs/guide.



  • @Tom-Elliott I’ll post the gen1\gen5\Optiplex7010 directory listing(s) but it’s affecting ALL images. All deploys fail with 1.3.X. This isn’t some system I casually use from time-to-time. This is a production system with storage nodes in NY, Dublin, Zurich and Singapore. These images are battle-tested and have been deployed hundreds of times each. That said…:

    Optiplex 7010
    Image path: /images/O7010V3
    Dir listing:
    -rwxrwxrwx 1 fog root 2 Oct 12 14:42 d1.fixed_size_partitions
    -rwxrwxrwx 1 fog root 15 Oct 12 14:42 d1.original.fstypes
    -rwxrwxrwx 1 fog root 259 Oct 12 14:42 d1.original.partitions
    -rwxrwxrwx 1 fog root 0 Oct 12 14:42 d1.original.swapuuids
    -rwxrwxrwx 1 fog root 8822083 Oct 12 14:45 rec.img.000
    -rwxrwxrwx 1 fog root 18137848997 Oct 12 14:59 sys.img.000

    Lenovo X1 (non-carbon, AKA ‘Gen1’)
    Image path: /images/X1CG1V2
    Dir listing:
    -rwxrwxrwx 1 fog root 2 Oct 11 14:00 d1.fixed_size_partitions
    -rwxrwxrwx 1 fog root 15 Oct 11 14:00 d1.original.fstypes
    -rwxrwxrwx 1 fog root 259 Oct 11 14:00 d1.original.partitions
    -rwxrwxrwx 1 fog root 0 Oct 11 14:00 d1.original.swapuuids
    -rwxrwxrwx 1 fog root 8686938 Oct 11 14:01 rec.img.000
    -rwxrwxrwx 1 fog root 16026941814 Oct 11 14:33 sys.img.000

    Lenovo X1 Carbon (NVMe/AKA ‘Gen5’)
    note: This is using a Kaby Lake Intel chipset. This chipset is only a minor iteration on the Skylake chipset which the Gen4 (Gen4 is also NVMe) used. For this reason, I’m attempting to write the ‘Gen4’ image to the ‘Gen5’

    Image path: /images/X1CG4V3
    Dir listing:
    -rwxrwxrwx 1 fog root 2 Oct 12 12:24 d1.fixed_size_partitions
    -rwxrwxrwx 1 fog root 15 Oct 12 12:24 d1.original.fstypes
    -rwxrwxrwx 1 fog root 259 Oct 12 12:24 d1.original.partitions
    -rwxrwxrwx 1 fog root 0 Oct 12 12:24 d1.original.swapuuids
    -rwxrwxrwx 1 fog root 8822643 Oct 12 12:25 rec.img.000
    -rwxrwxrwx 1 fog root 15994261402 Oct 12 12:52 sys.img.000


  • Senior Developer

    I see what the problem is.

    What I don’t understand is what the OS is.

    Can you get contents of ls /images/<relevantimagehere>?


  • Senior Developer

    can you try:
    parted -s /dev/nvme0n1 mkpart primary ntfs 206848s 100%


  • Senior Developer

    @BardWood It’s a 512 Gb disk?



  • @Tom-Elliott Tried that (see below): Output = ‘Error: You requested a partition from 106mb to 512gb (sectors 206848… 1000215215)’


  • Senior Developer

    @BardWood That message (the it exists is non-issue).

    SO now that we know the device name (it’s not /dev/sda)

    Try the fog until it fails.

    Then when it fails at the prompt try:

    parted -s /dev/nvme0n1 mkpart primary ntfs 206848s -- -1s



  • @Tom-Elliott Additional info from Optiplex Desktop:

    0_1489609827234_fog-o7010.jpg



  • @Tom-Elliott Results of your suggestions (on Gen5/NVMe):

    parted -s /dev/nvmen0 mkpart primary ntfs 206848s – -1s
    ‘Could not stat device /dev/nvmen0 - No such file or directory’

    So I re-ran with a slight syntax change based on lsblk output
    Input:
    ‘parted -s /dev/nvmen0n1 mkpart primary ntfs 206848s – -1s’
    Output:
    Error: You requested a partition from 106mb to 512gb
    (sectors 206848… 1000215215)
    The closest location we can manage is 512GB to 512GB (sectors 1000212480… 1000215215.
    So, at least parted sees it now. Small victories.

    Results of lsblk -dpn (-dpno spit out the help file. -d/-dp/-dpn had similar output. ‘-o’ either on its own or in combo with other switches resulted in help file).
    0_1489609197631_gen5-lsblk-dpn.jpg

    Results of ‘fdisk -l’
    0_1489609244337_gen5-fdisk-l.jpg


  • Senior Developer

    @BardWood The debug, because you have nvme, probably needs:

    /dev/nvmen0 (a fdisk -l or lsblk -dpno might prove more fruitful here)



  • @Tom-Elliott No, that was from the X1 ‘Gen1’. The NVMe machine (Gen5) error is: X1 Carbon, Gen 5:
    Error: Could not stat device /dev/sda - no such file or directory

    FYI: I updated to RC-15 but so far no change in the errors. I’ve only tested RC-15 on the ‘Gen1’. I’ll try to re-run the parted command on the Gen5.


  • Senior Developer

    @BardWood The message of the “Output of above command: ’ Error: The location 206848 is outside of the device /dev/sda’” Was this on the machine with the NVMe disk?



  • @Tom-Elliott I saw that in the screenshot, ‘recovery partition’. Thing is, I booted that machine off a Gparted live USB image (latest version as of today) and it only sees 2. The MS riser and primary (technically, both are primary but you know what I mean). I’m trying to deploy a Win7, full disk, multi-partition. No recovery partition spotted so far. The disk on the ‘Gen5’ is using NVMe on SATA3(6gb).


  • Senior Developer

    @BardWood Not related.

    Is this image a single partition image? Or do you have a recovery partition as well?



  • @Tom-Elliott

    X1 Carbon, gen1:
    Output of above command:
    ‘Error: The location 206848 is outside of the device /dev/sda’
    *This only has a 160G disk so 206848 may be too large.

    Optiplex 7010, 500GB disk:
    Error: Can’t have overlapping partitions.

    X1 Carbon, Gen 5:
    Error: Could not stat device /dev/sda - no such file or directory

    Unsure if this is related but while FOG is booting the kernel, I see output stating:

    “Populating /dev using udev: udevd[2536]: error creating epoll fd: Function not implemented”



  • PS. Same result as in my screenshot on our Dell Optiplex 7010 desktop.



  • @Tom-Elliott said in Please point me to 1.2.0 -> 1.3.4 for Centos 6.7 upgrade docs/guide.:

    parted -s /dev/sda mkpart primary ntfs 206848s – -1s

    Output of above command:
    ‘Error: The location 206848 is outside of the device /dev/sda’



  • @BardWood Just saw your message below. I’ll attempt debug.



  • @Tom-Elliott I will double-check the block size on the ‘Gen 5’. That’s a solid lead. In the case of the ‘Gen 1’ master and attempting to write the image that came from it back to it, there is no change whatsoever. I’ll attempt to image a few more machines of different types (I’ll attempt one of our desktops, Dell Optiplex 7010). Both the Optiplex and Lenovo X1 ‘Gen 1’ images worked with 1.2.0 and I wrote both of those images back to their respective machine types last week. Is there a way to gather additional data? I see 'Log level 4" in the parameters but found no corresponding data on the Web interface.


  • Senior Developer

    Further looking over, it appears the “resize” code isn’t even to blame, but rather the parted command is unable to create the partition.

    You could try running a debug session.

    When it fails to run, try running:

    parted -s /dev/sda mkpart primary ntfs 206848s -- -1s


  • Senior Developer

    @BardWood I tested this and had no issues. I’m not sure where the disconnect is, but my guess is the partition layout is not able to be placed on that system.

    You can say otherwise, my testing took an image from 1.2.0 captured, tested on three different systems.

    Something is different somewhere. I am most sure of it. Now there is the possibility your disk is now a 4k disk? I’m only guessing based on the information I have. I assure you, the legacy placement is working as intended. The only difference I can think of is the way resizing occurred in 1.2.0 vs how it occurs in 1.3.x now. These changes were needed and are unlikely to change.

    I’m not lying however, that the issue (even if it is as I suspect due to 4k vs 512 bytes) is because one disk size is different from the other.


Log in to reply
 

373
Online

6.0k
Users

13.3k
Topics

125.2k
Posts