Issues with migrating CentOS from ESXi to Proxmox KVM

[moonjazz]

Hi all,

I am happy to announce I finally pulled the trigger and am moving away from ESXi to Proxmox. I started the process last night and everything was going swimmingly(I followed the instructions here: https://pve.proxmox.com/wiki/Migration_of_servers_to_Proxmox_VE#VMware_to_Proxmox_VE_.28KVM.29). All of my Windows and Ubuntu virtual machines booted fine and worked right out of the box, no issues. However, the bulk of my virtual machines are CentOS 7 and with every single one of them, they boot directly into the dracut emergency shell. I have 2 pictures attached of the error messages that are thrown in the dracut shell.

Hopefully someone can help me out so I can finally use Proxmox instead of ESXi.

Thanks,
Ryan

 

[udo]

Hi,
what kind of driver do you use for your centos-disk? virtio, scsi, ide?
You can try to use ide, or use an centos-cdrom for rescue-mode (I assume the drive-driver is missing in the initramfs).

Udo

 

[moonjazz]

Hi,
what kind of driver do you use for your centos-disk? virtio, scsi, ide?
You can try to use ide, or use an centos-cdrom for rescue-mode (I assume the drive-driver is missing in the initramfs).

Udo

I'm new so I just left it as the default which was VIRTIO. I'll try out SCSI and then IDE and report back.

 

[moonjazz]

I was unable to boot into the OS when using IDE or SCSI, it booted straight to PXE.

 

[fireon]

Had similar problems with Centos migrating from PHY to virtual. I've installed mainlinekernel from repo, that solved the problem immediately. Then you can easy setup hotplug too. Done here with Centos6 and 7, both with scsi-bus. Dont forget to change UUID in fstab and write grub.

 

[moonjazz]

Had similar problems with Centos migrating from PHY to virtual. I've installed mainlinekernel from repo, that solved the problem immediately. Then you can easy setup hotplug too. Done here with Centos6 and 7, both with scsi-bus. Dont forget to change UUID in fstab and write grub.

Hmmm that seemed to fix the issue, however when I installed the mainline kernel from the rescue kernel, all of my data from after the rescue kernel was gone. Anyway around that?

 

[fireon]

all of my data from after the rescue kernel was gone

What do you mean with that, what data?

 

[moonjazz]

What do you mean with that, what data?

Sorry, I read that over myself and it was kind of confusing. The only was I was able to install the mainline kernel as you mentioned, was when I booted into the rescue kernel. From there I installed the ml-kernel package. I rebooted the virtual machine and everything booted fine. On that virtual machine I was running a few services and databases and it seems that they had reverted to a previous state from a few months back, missing new entries. I have no clue why that would be.

 

[udo]

Sorry, I read that over myself and it was kind of confusing. The only was I was able to install the mainline kernel as you mentioned, was when I booted into the rescue kernel. From there I installed the ml-kernel package. I rebooted the virtual machine and everything booted fine. On that virtual machine I was running a few services and databases and it seems that they had reverted to a previous state from a few months back, missing new entries. I have no clue why that would be.

Hi,
that sounds, that your used harddisk was transfered "months back".

You are sure, that you use the latest version of the hdd? (fresh converted from esxi)

Udo

 

[moonjazz]

Hi,
that sounds, that your used harddisk was transfered "months back".

You are sure, that you use the latest version of the hdd? (fresh converted from esxi)

Udo

Yup, it seems as if that's my issue. I'm checking the VMs VMDKs and it seems there are multiples. For instance there's "VMname.vmdk" and "VMname-000001.vmdk" and "VMname-000002.vmdk". Usually I just convert the flat VMDK but it seems as if it's the old virtual disk. How would I convert the non-flat VMDK?

 

[9acca9]

Had similar problems with Centos migrating from PHY to virtual. I've installed mainlinekernel from repo, that solved the problem immediately. Then you can easy setup hotplug too. Done here with Centos6 and 7, both with scsi-bus. Dont forget to change UUID in fstab and write grub.

Hi. Im having the same problem here. Can you give me a hand, on how to do that???
Thanks

 

[jeremyjohnson]

I know this is an old thread but I just went through this exact issue and found a solution for myself.

I tried the ML kernel but that didn't work. I ended up rebuilding the initramfs and then the server would boot. Once that was done, my NIC names and UUID changed so I got the new UUID and reconfigured my network-scripts file with the correct UUID and it is working. This is on a DNS server and I immediately saw it start answering DNS queries.

I am not migrating my other DNS server and will use the ML kernel so they match but I don't think I needed it. I'll be doing other VMs later and will see if I can get them to work without the new kernel.

Update: I just tested on my grafana server and was able to get it working by only updating the initramfs. I realize now when I tried earlier, there was a mixup on what uname -r gave me vs what the default kernel I was booting from. Once I straigtened that out, it was a very simply process to just pdate initramfs and then the NIC UUID.

 

[rhael79]

Cambia el Disco a IDE y tu tema sera solucionado magicamente