ESXi Import Wizard Error

[pstorkey]

I am working on moving from a single free instance of ESXi to a single PVE server, using the same hardware.
As an interim step, I have installed PVE as a VM on the ESXi instance, and I am using a NFS share to import the VMs. The plan is to:

• Import all VMs from ESXi to the NFS volume, using the import utility in the PVE instance (VMDK converted to qcow2)
• Blow away ESXi from the physical hardware, and install PVE on the same hardware
• Import the qcow2 VMs into the new PVE instance on the physical hardware

I have successfully imported a number of Linux and Windows VMs to the NFS without instance, but have run into a problem with a Windows file server with a 1TB data drive. The 100GB system drive imports successfully, but the 1TB data drive fails at 53% every time. It is a thin provisioned VMDK with 1 TB GPT NTFS volume, with approximately 300GB free.


The import utility fails with the following error:

qemu-img: error while reading at byte 591161842176: Input/output error
TASK ERROR: unable to create VM 107 - cannot import from 'esxi:ha-datacenter/HDD Datastore 1/smeg-files.smeghead.ca/smeg-files.smeghead.ca_1.vmdk' - copy failed: command '/usr/bin/qemu-img convert -p -n -f vmdk -O qcow2 '/run/pve/import/esxi/esxi/mnt/ha-datacenter/HDD Datastore 1/smeg-files.smeghead.ca/smeg-files.smeghead.ca_1.vmdk' zeroinit:/mnt/pve/nas/images/107/vm-107-disk-1.qcow2' failed: exit code 1

The Linux journal shows the following error:



pve esxi-folder-fuse[1083]: error handling request: cached read failed: error reading a body from connection: end of file before message length reached
                                           
                                           Caused by:
                                               end of file before message length reached
pvedaemon[566872]: VM 107 creating disks failed
pvedaemon[566872]: unable to create VM 107 - cannot import from 'esxi:ha-datacenter/HDD Datastore 1/smeg-files.smeghead.ca/smeg-files.smeghead.ca_1.vmdk' - copy failed: command '/usr/bin/qemu-img convert -p -n -f vmdk -O qcow2 '/run/pve/import/esxi/esxi/mnt/ha-datacenter/HDD Datastore 1/smeg-files.smeghead.ca/smeg-files.smeghead.ca_1.vmdk' zeroinit:/mnt/pve/nas/images/107/vm-107-disk-1.qcow2' failed: exit code 1

I have repeated the import several times, and it always fails at exactly the same point. I have tried verifying the VMDK, and checking the NTFS filesystem, and both look good.

ESXi host is version 6.7 Update 3, PVE version is 8.2.



Note, I have purchased a community subscription, but have not applied it to the temporary PVE VM. I will apply the key to the final hardware PVE instance.



Any help would be most appreciated.

 

[Zerstoiber]

I think you hit a bug with spaces in the filename/path ("HDD Datastore 1")

The Patchseries is here,
https://www.mail-archive.com/pve-devel@lists.proxmox.com/msg19354.html

but it seems not to be released yet:
https://git.proxmox.com/?p=pve-esxi-import-tools.git;a=log

 

[Zerstoiber]

Oh, just noticed, the fix is already located within PVE storage (not within pve-esxi-import-tools):
https://git.proxmox.com/?p=pve-storage.git;a=log
https://git.proxmox.com/?p=pve-storage.git;a=commit;h=22cb4b0b7812658ecc2444ca3ee51a04c50e4f57

You'll have to wait for it to receive the next version bump and then update your Proxmox nodes accordingly

 

[pstorkey]

Oh, just noticed, the fix is already located within PVE storage (not within pve-esxi-import-tools):
https://git.proxmox.com/?p=pve-storage.git;a=log
https://git.proxmox.com/?p=pve-storage.git;a=commit;h=22cb4b0b7812658ecc2444ca3ee51a04c50e4f57

You'll have to wait for it to receive the next version bump and then update your Proxmox nodes accordingly

Thanks for the reply, Zerstoiber. I tried renaming the datastore to "HDD_Datastore_1", but unfortunately it still failed in the same spot. Do you think the patch would work?

 

[pstorkey]

I ended up detaching the VMDK, cloning it to another datastore, and reattaching it. I'm happy to say, this allowed the import to complete successfully.