How to switch CentOS8 VM to use a SCSI or SATA Controller for the booting vHDD instead of a NVME?

[swinster]

Hi,

We have a CentSO 8 VM which was built using an NVME Controller for the vHDD. Due to some special needs, we must change this to a SCSI Controller. However, if we add the SCSI Controller to the VM hardware and then assign the vHDD to use the control, the OS fails to boot correctly. 

From https://docs.vmware.com/en/VMware-vSphere/7.0/com.vmware.vsphere.vm_admin.doc/GUID-5872D173-A076-42F... there is a "Caution" note that states the driver need to be in the OS. We can boot the VM with another vHDD attached to the SCSI controller with no issue, and we can see the drive within the OS, so the driver must be available.; however, switching the primary boot drive to use the SCSI control still fails, so I guess there are some Linux config files that need to be adjusted (if possible).

Has anyone got any suggestions on how this might be resolved?

Cheers.

[swinster]

FWIW, I have tried several options. I can use and IDE controller. The VM will boot normally, and I can remove the NVME, but as soon as I add a SCSI controller and set the vHDD to use SCSI, I get the same issue - the CentOS VM will boot into emergency mode (dracut).

I have also tried to use VMware converter, but this doesn't;t seem to make a difference. I thought I could downgrade the virtual hardware version when converting (the current version is 14), but setting anything lower than 14 results in an "invalid configuration for device 0". Still, I'm not sure if this would have resolved the issue.

I'm currently running another conversion and have set the controller in the "Other" section of the setting to be SCSI, so I'll see what happens here, but I am not hopeful.

[swinster]

Unfortunately (but as suspected), the conversion to use a SCSI controller also did not work and the VM boots into emergency mode. 

[swinster · Accepted solution]

OK, someone on the CentOS forums managed to chime in and help.

They suggested booting into the rescue kernel (which was the last one on the list during the initial boot, then running either of:

yum update to the latest kernel 

or

yum reinstall kernel-$your-latest-kernel-version

or

dracut -f --kver=$your-latest-kernel-version

where `$your-latest-kernel-version`  was the highest version number listed in

ls /lib/modules

I ran the first `yum` update, but whilst a few packages were updated, it did not resolve the issue. I then when straight to the `dracut` command, which did resolve the issue. 

Nice.