CentOS VM won't boot after migration from VMware

[JohnOCFII]

Greetings! New Proxmox user here. Newly installed VMs are working fine and Debian VMs migrated from VMware ESXi 8.x are working well, but, I've having trouble migrating a couple of older CentOS VMs. The VMs were exported from the ESXi web client export option that creates OVF output.

I've attempted to import from OVF with qm importovf 112 CentWatch.ovf local-zfs.
The VM starts, but apparently doesn't find the disks. I did attempt to switch to the SATA or even IDE disk by using the detach feature but that didn't change anything.

The issue feels very similar to that experienced in this thread: https://forum.proxmox.com/threads/ancient-centos-5-vm-wont-boot-after-transfer.136845/

I've included zipped versions of the OVF file exported from VMware and the created .conf file.

Looking for any advice.

Thanks!

John

 

[jsterr]

Greetings! New Proxmox user here. Newly installed VMs are working fine and Debian VMs migrated from VMware ESXi 8.x are working well, but, I've having trouble migrating a couple of older CentOS VMs. The VMs were exported from the ESXi web client export option that creates OVF output.

I've attempted to import from OVF with qm importovf 112 CentWatch.ovf local-zfs.
The VM starts, but apparently doesn't find the disks. I did attempt to switch to the SATA or even IDE disk by using the detach feature but that didn't change anything.

The issue feels very similar to that experienced in this thread: https://forum.proxmox.com/threads/ancient-centos-5-vm-wont-boot-after-transfer.136845/

I've included zipped versions of the OVF file exported from VMware and the created .conf file.

Looking for any advice.

Thanks!

John

Hello. I always use https://www.thomas-krenn.com/de/wiki/VM_Export_&_Import:_Migration_von_Windows_/_Linux_VMs_von_VMware_zu_Proxmox_VE how does your exported files look like on pve?

for example:

oot@pve-01:~/VMS# cd JS-AD-01
root@pve-01:~/VMS/JS-AD-01# ls
JS-AD-01-disk1.vmdk  JS-AD-01-file1.nvram  JS-AD-01.mf  JS-AD-01.ovf

Whats the error on startup? Your qm-config file also looks short, should look like more then this. It looks like your missing a scsicontroller (for example scsihw:virtio-scsi-single

root@PMX5:~# cat /etc/pve/qemu-server/112.conf
boot: c
bootdisk: scsi0
cores: 4
cpu: cputype=host
memory: 8192
name: test5
ostype: l26
scsi0: vm_nvme:vm-112-disk-0,cache=writeback,discard=on,size=80000M,ssd=1
scsihw: virtio-scsi-single
serial0: socket
smbios1: uuid=00ed952f-5db8-44a7-a5e4-e250d7a697b8
vga: serial0
vmgenid: 441f09c8-ea02-482c-b2d9-e5505b55ccf6

 

[LnxBil]

The VM starts, but apparently doesn't find the disks.

In Grub or later? Normally migrating Linux VMs is a breeze.

 

[JohnOCFII]

In Grub or later? Normally migrating Linux VMs is a breeze.

It hangs here. And yeah, my Debian VMs moved easily.

 

[bbgeek17]

Have you tried to change Bios to UEFI?
Did you set the Boot order in VM config? Note that "bootdisk" is depreciated:

--bootdisk (ide|sata|scsi|virtio)\d+
           Enable booting from specified disk. Deprecated: Use boot: order=foo;bar instead.

You can also try to boot from ISO into a rescue mode and troubleshoot there, ie can you see the disk, what is on the disk, can you mount it, etc. It'd be good confirm you have actual data on them.

Good luck

---

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox

 

[JohnOCFII]

Whats the error on startup?

Missed responding to this one:

I get the grub message about Booting from Hard Disk... and then it just hangs.

 

[JohnOCFII]

Have you tried to change Bios to UEFI?

I did try that, and got as far as waiting for it to timeout at the #dracut prompt, but there is no /boot directory, so I wasn't sure where to go from there. I had started to look at this reference: https://www.cjcheema.com/2019/06/17/how-to-recover-or-rebuild-initramfs-in-centos-7-linux/

Did you set the Boot order in VM config? Note that "bootdisk" is depreciated:

--bootdisk (ide|sata|scsi|virtio)\d+
           Enable booting from specified disk. Deprecated: Use boot: order=foo;bar instead.

I did check the boot order. Right now, it the only activated entry is scsi I've tried IDE and SATA as well. No change in behavior. I'll poke around a bit, then blow it away, and re-do the qm import and try again.

You can also try to boot from ISO into a rescue mode and troubleshoot there, ie can you see the disk, what is on the disk, can you mount it, etc. It'd be good confirm you have actual data on them.

Yes, I wanted to pursue this. Being new to Proxmox, I wasn't sure of the best way to do this. Do I start with a rescue ISO and install that as a VM and then somehow move the created VM image over to the directory with the Rescue ISO? I'm still learning the logistics of this environment compared to ESXi.

Thanks!

 

[bbgeek17]

Yes, I wanted to pursue this. Being new to Proxmox, I wasn't sure of the best way to do this. Do I start with a rescue ISO and install that as a VM and then somehow move the created VM image over to the directory with the Rescue ISO? I'm still learning the logistics of this environment compared to ESXi.

Just pick any Linux ISO, whether dedicated rescue or regular install, attach to VM and set Bios to boot from it. Practically all ISOs allow you do drop down to a shell and run "lsblk" and "lsscsi" , as well as "fdisk" and "mount".

Alternatively, you can create new VM, install fresh OS, then import your old VM as extra disk and have full OS toolset at your disposal for troubleshooting. This route is more to confirm that you actually have data there, to avoid chasing troubleshooting boot from a corrupted/bad disk.

---

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox

 

[JohnOCFII]

Just pick any Linux ISO, whether dedicated rescue or regular install, attach to VM and set Bios to boot from it. Practically all ISOs allow you do drop down to a shell and run "lsblk" and "lsscsi" , as well as "fdisk" and "mount".

Thanks for the guidance. Seems I can't cut and paste from the Proxmox console window. Here's some output. Disk is there... but I think the default LVM from the CentOS installation is related to my issue:

 

[LnxBil]

So, the layout is UEFI. Please change to UEFI and report back the actual error. I have not seen any screenshot with UEFI of the error so far. The actual message vom dracut would be interessting.

 

[bbgeek17]

@LnxBil is correct, the VM was installed with UEFI and needs to be configured as such. We also learned that data/partitions/layout is there, so you are not trying to boot from empty disk.

There is no indication that CentOS layout is contributing to your problem. You attempted to mount a physical device which does not contain any filesystem. Dont forget that the physical disk is wrapped into LVM physical volume, which is wrapped into LVM volume group, which is subdivided into LVM local volumes, which actually contain the filesystem that "mount" can handle.

Plug this into google "troubleshoot uefi boot virtual machine", even guides from Microsoft could be helpful:
https://learn.microsoft.com/en-us/t...al-machines/azure-linux-vm-uefi-boot-failures

Good luck

---

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox

 

[JohnOCFII]

Thank you all for your guidance, especially @bbgeek17, @LnxBil, and @jsterr!

The migrated CentOS VM is booting correctly now!

I had previously tried UEFI boot, but I think I had mistakenly changed some other settings regarding disk or controller type at that time. This time, it all worked properly. Here's the ending 112.conf file.

Thanks again!

John

bios: ovmf
boot: order=sata0
cores: 4
efidisk0: local-zfs:vm-112-disk-1,efitype=4m,pre-enrolled-keys=1,size=1M
machine: q35
memory: 4096
name: CentWatch
net0: virtio=BC:24:11:07:CD:DD,bridge=vmbr0,firewall=1
sata0: local-zfs:vm-112-disk-0,size=48G
scsi1: none,media=cdrom
scsihw: virtio-scsi-pci
smbios1: uuid=86496eec-2e51-4fc3-b1e4-8eac53203b55
vmgenid: 73b66a52-e458-4a72-ab8e-cd54250db76a