New Import Wizard Available for Migrating VMware ESXi Based Virtual Machines

[flp]

Hi there,

I got this error message trying to import an Ubuntu VM from ESXi v7. Seems the script doesn't like the name or path of the esxi files. Any suggestion anybody? VM is power off and I'm just using default settings in the migration wizard.

Use of uninitialized value $parent in concatenation (.) or string at /usr/share/perl5/PVE/Storage/Plugin.pm line 985.
TASK ERROR: unable to create VM 101 - cannot import from 'Bravo_ESXI:ha-datacenter/Group2B/ODOO Ubuntu Server 22.04.3 Base-replica/ODOO Ubuntu Server 22.04.3 Base-replica-000001.vmdk' - parent '' includes whitespace

 

[Ramalama]

Hi there,

I got this error message trying to import an Ubuntu VM from ESXi v7. Seems the script doesn't like the name or path of the esxi files. Any suggestion anybody? VM is power off and I'm just using default settings in the migration wizard.

Use of uninitialized value $parent in concatenation (.) or string at /usr/share/perl5/PVE/Storage/Plugin.pm line 985.
TASK ERROR: unable to create VM 101 - cannot import from 'Bravo_ESXI:ha-datacenter/Group2B/ODOO Ubuntu Server 22.04.3 Base-replica/ODOO Ubuntu Server 22.04.3 Base-replica-000001.vmdk' - parent '' includes whitespace

I have the same issue on one VM, the weird thing is, that i migrated already some VM's with Whitespaces.
Migrated 20 VM's now, one has this whitespace issue...

EDIT:
I found an Answer to my Performance issue, If i do 5 Import Tasks at same time, all 5 import tasks Combined are like 4x faster as only one import task.
Im not sure if it has something todo with multithreading of some sort? Or whatever, but thats how i imported overnight like 10 VM's or 20 over the weekend.

Cheers

 

[Ramalama]

Same happens if i Migrate VMs, just found out,

If i Migrate only one, i get max 6-7 Gibps.
But if i migrate 3 at the same time, you see in the picture, im getting 16-17Gibps.

And its somehow not the CPU's, nothing reaches 100%, max 20-30%, its a high Clocked Genoa 9374F in both servers.
There is something with multithreading or network threading, but the Intel E810 is set to 64 Qeues, so each cpu has its own. It's weird, something is wrong and i dont get what.

Same thing with the Import Wizard, but there is one Connection limited to around 1GIbps, 3 imports at same time get like 3Gibps.
I cannot test more imports sadly anymore, cause i imported already all VM's.
And it has nothing todo with Lacp etc, a single iperf3 test with an mtu of 1500 does 24 Gibps between the Servers, multiple Iperf Instances do around 40-45 Gibps.

If anyone of the Proxmox Team wants to debug this with me somehow, or need sth, let me know. Cheers

 

[flp]

I have the same issue on one VM, the weird thing is, that i migrated already some VM's with Whitespaces.
Migrated 20 VM's now, one has this whitespace issue...

EDIT:
I found an Answer to my Performance issue, If i do 5 Import Tasks at same time, all 5 import tasks Combined are like 4x faster as only one import task.
Im not sure if it has something todo with multithreading of some sort? Or whatever, but thats how i imported overnight like 10 VM's or 20 over the weekend.

Cheers

Have you managed to migrate it? we have around 50 VMs to migrate with similar name format containing white spaces. I hope Proxmox devs give us an answer to this bug cause changing the folder names is not an option.

I messed around with the perl file comenting the lines but, as expected, the system didn't buy it and refused to reboot. Have to ssh and undo the editing. Anyway we'll have to wait a bit longer...

 

[Ramalama]

Have you managed to migrate it? we have around 50 VMs to migrate with similar name format containing white spaces. I hope Proxmox devs give us an answer to this bug cause changing the folder names is not an option.

I messed around with the perl file comenting the lines but, as expected, the system didn't buy it and refused to reboot. Have to ssh and undo the editing. Anyway we'll have to wait a bit longer...

I will find out pretty fast, just didnt had time, its one VM here, that no one needs anyway, so i can experiment with it.
It just confuses me, because the error itself looks missleading to me, like i said, i migrated like 20 VM's and at least 5 had Whitespaces and i had no issues, just one doesn't want and has this whitespace issue.

Glad god it is an VM that no one needs xD
So i was lucky, but im going to experiment in a bit, just have to finish everything else....

Migration is a time consuming task, especially with Windows 7 and Windows Server 2012...
That old Crap VMs doesn't work reliably with ballooning, everything that is newer (10/2016/2019/2022) is perfect.
Had even to upgrade Centos 6/7 Vms here with newer kernels, because of some issues. I didn't expected that Centos 7 runs absolutely amazing with elrepo's kernel 6.8, lol & CentOS 6 with 4.4 Kernel got new life xD

However its all offtopic, all i can say the migration Process took very long, vmware tools uninstall doesnt work and upgrades and reactivating windows with some blocked licences and so on...
But the Proxmox part itself (Migration Wizard) and overall everything else worked flawless, except that one VM xD

However, gonna check now how to fix it, but in the last rescue, there is the ovftool, which works very good either!
I mean you can still go that route with ovftool, its not like its a lot harder or something.

Cheers

 

[Ramalama]

ive ended up simply cloning the vm to another name on esxi and then migrated.
Its not suitable for a lot of vms, but for only one (my situation) it was the most comfortable way.

 

[PLP-Lack]

If i import VMs from ESXI, will the "UUID" stay the same so that all service and licences work like before?

 

[Ramalama]

If i import VMs from ESXI, will the "UUID" stay the same so that all service and licences work like before?

depends, on windows 10/2016/2019/2022 definitively yes, basically no matter what happens, they keep the license, so no worries.

Windows 7 and W2012/R2, määh, some of them lost here licenses, but not all, so i don't know from what it depends.
That ones that lost the licence, had the issue that i couldnt reactivate them with the same licence again, because they all were blocked by microsoft, whyever...
So i had to get new licences for them, we have luckily some left.

Cheers

 

[PLP-Lack]

depends, on windows 10/2016/2019/2022 definitively yes, basically no matter what happens, they keep the license, so no worries.

Windows 7 and W2012/R2, määh, some of them lost here licenses, but not all, so i don't know from what it depends.
That ones that lost the licence, had the issue that i couldnt reactivate them with the same licence again, because they all were blocked by microsoft, whyever...
So i had to get new licences for them, we have luckily some left.

Cheers

Some years ago i unsed the VMWare Migration tool to migrate a VM from HyperV and with this tool it was not possible to keep the UUID.
Then some services from my 2019 Server where not working any more. So i used Veem Backup to the the migration.

This is the reason i ask, because i don't want to get the same troubles.

 

[fiona]

Hi,

Hi there,

I got this error message trying to import an Ubuntu VM from ESXi v7. Seems the script doesn't like the name or path of the esxi files. Any suggestion anybody? VM is power off and I'm just using default settings in the migration wizard.

Use of uninitialized value $parent in concatenation (.) or string at /usr/share/perl5/PVE/Storage/Plugin.pm line 985.
TASK ERROR: unable to create VM 101 - cannot import from 'Bravo_ESXI:ha-datacenter/Group2B/ODOO Ubuntu Server 22.04.3 Base-replica/ODOO Ubuntu Server 22.04.3 Base-replica-000001.vmdk' - parent '' includes whitespace

FYI, a fix for this has been applied in git and is expected to be in the next version of libpve-storage-perl (will most likely be 8.2.2).

 

[flp]

Hi,

FYI, a fix for this has been applied in git and is expected to be in the next version of libpve-storage-perl (will most likely be 8.2.2).

Thanks Fiona, looking forward to it!

 

[sockmunky]

This is great timing for my planned move off of vCenter / ESXI!

I was running into an error repeatedly on any VM I tried live migrating:

qemu-img: error while reading at byte 0: Input/output error
TASK ERROR: unable to create VM 104 - cannot import from 'esxi:ha-datacenter/Local/VM Test/VM Test.vmdk' - copy failed: command '/usr/bin/qemu-img convert -p -n -t none -f vmdk -O raw /run/pve/import/esxi/esxi/mnt/ha-datacenter/Local/VM Test/VM Test.vmdk zeroinit:/dev/zvol/Datastore02/vm-104-disk-0' failed: exit code 1

I had success after going through full VM prep and shutting the machine down, I just couldn't get live migrate to work. But that's okay, still worked great for offline migration.

Thanks!

 

[Azunai333]

I just couldn't get live migrate to work

You always have to shutdown the VM on the ESXI.
The online migration will start the VM on the PVE right away. The migration of the virtual disks will happen in the background.

 

[stv.Bundesadmin]

yeah, "Live Import" is a bit misleading just like Tesla Autopilot ;-)

i think it was mentioned already that it would make sense to actually check via API for the VMware VM to be really shut down.

 

[z0ddein1]

I am having an issue where I see

failed to spawn fuse mount, process exited with status 65280 (500)

when I try to create the ESXi import storage.
The target host is a ESXi 6.7 (Image profile: (Updated) ESXi-7.0U1c-17325551-standard (VMware, Inc.))
I have already set the 2 <soap> xml tags and verified it is saved on disk.

I have also rebooted both hosts (ESXi and Proxmox) multiple times.
Am I missing something and is there any workaround or fix for my issue?

Update: It was due to having the same name for 2 different type of datastore on the ESXi, in my case it was a NFS and a iSCSI datastore having the same name (for some reason it allowed it to exist).
I just disconnected the unused one then it is able to show the list of VMs I have on the ESXi host.

However now I have another issue where the VM disk names containing non-ASCII characters and seems like it is not able to find the files.

 

[ronejamesdo]

I am having an issue where I see

failed to spawn fuse mount, process exited with status 65280 (500)

when I try to create the ESXi import storage.
The target host is a ESXi 6.7 (Image profile: (Updated) ESXi-7.0U1c-17325551-standard (VMware, Inc.))
I have already set the 2 <soap> xml tags and verified it is saved on disk.

I have also rebooted both hosts (ESXi and Proxmox) multiple times.
Am I missing something and is there any workaround or fix for my issue?

Update: It was due to having the same name for 2 different type of datastore on the ESXi, in my case it was a NFS and a iSCSI datastore having the same name (for some reason it allowed it to exist).
I just disconnected the unused one then it is able to show the list of VMs I have on the ESXi host.

However now I have another issue where the VM disk names containing non-ASCII characters and seems like it is not able to find the files.

I had the same error (and had tried the <soap> thing too). Your description lead me to think that this error might just be an obscure reaction to the wizard encountering some ESXi Datastore condition it could not parse rightly.

My ESXi Datastore was called "SSD Raid6 + 2 spares" and renaming it to "SSD Raid6 and 2 spares" fixed it for me.

Apparently the wizard does not like special characters (at least the plus sign) in Datastore names, which is sensible, and renaming the Datastore is done in the ESXi GUI and just takes like one second.

Thanks.

 

[flp]

We keep getting "Connection error - Timeout" when adding the esxi datastorage of a remote esxi host in our datacentre from a pve in our office LAN.

The link is not that slow, it pings at 10 to 30ms. We connect ok to the remote esxi gui via browser and we can add that esxi storage ok from another pve host in the same subnet at the datacentre so the esxi can be connected ok with a proxmox for migration but still times out when trying over the office end of the mpls link. Any ideas?

Thanks!

 

[masagun]

this good news for me... tanks for update fitur...

 

[jlauro]

We keep getting "Connection error - Timeout" when adding the esxi datastorage of a remote esxi host in our datacentre from a pve in our office LAN.

The link is not that slow, it pings at 10 to 30ms. We connect ok to the remote esxi gui via browser and we can add that esxi storage ok from another pve host in the same subnet at the datacentre so the esxi can be connected ok with a proxmox for migration but still times out when trying over the office end of the mpls link. Any ideas?

Thanks!

IMO, an extra 10-30ms is on the slow side for a virtual disk access...
I recommend doing a transfer of the data so it's local and then manually importing the vm without the wizard.
That said, I'm surprised it's timing out. I would expect it to be painfully slow, but does seem like it should run.

 

[z0ddein1]

I had the same error (and had tried the <soap> thing too). Your description lead me to think that this error might just be an obscure reaction to the wizard encountering some ESXi Datastore condition it could not parse rightly.

My ESXi Datastore was called "SSD Raid6 + 2 spares" and renaming it to "SSD Raid6 and 2 spares" fixed it for me.

Apparently the wizard does not like special characters (at least the plus sign) in Datastore names, which is sensible, and renaming the Datastore is done in the ESXi GUI and just takes like one second.

Thanks.

Unfortunately it is not easy to rename the vdisk of a VM since it consists of multiple files and each file references other files internally. Hope the non-ascii issue get fixed but I have migrated my VMs using clonezilla manually in the end.