migration plan from vsan cluster to proxmox

Hello
to migrate from vSAN , i first put in place vsan files services and create an nfs share and mount it on esxi, then i shut vm and storage move to nfs share. On proxmox side i can see nfs share as well, i try to use qm import directly from nfs share
unable to parse volume ID './Clone_mavm.vmx'
using qm import with ESXI_FQDN:ha-datacenter/DS_migration_Proxmox/Clone_mavm/Clone_mavm.vmx works and copy all files

Is there any way to use qm import with an nfs share directly ?

if i use gui i can import vm without disks and in second step i can qm import disk from nfs share (largely faster against gui import), but as i need to bulk import hundreds of vm i try to find cli command to import only vm skel (without importing disks) and use qm import disk from nfs sources

Regards
 
Last edited:
i try to use qm import directly from nfs share
unable to parse volume ID './Clone_mavm.vmx'
using qm import with ESXI_FQDN:ha-datacenter/DS_migration_Proxmox/Clone_mavm/Clone_mavm.vmx works and copy all files

Is there any way to use qm import with an nfs share directly ?
Hi @Xavier Droniou , can you provide exact commands with subsequent output that you executed? As well as your storage configuration:
cat /etc/pve/storage.cfg

To answer your question: Yes, you can can import file from locally mounted path, generally you will need to provide absolute path to the file you are importing.

Cheers


Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox
 
Bash:
cat /etc/pve/storage.cfg
dir: local
        disable
        path /var/lib/vz
        content snippets
        shared 0

rbd: R3_Ceph_Datastore
        content images,rootdir
        krbd 0
        pool R3_Ceph_Datastore

cephfs: R3_Cephfs
        path /mnt/pve/R3_Cephfs
        content vztmpl,import,iso
        fs-nameR3_Cephfs

pbs: pool_zfs_01
        datastore pool_zfs_01
        server 1.2.7.1
        content backup
        fingerprint e1:d7:71:f3:bc:49:0b:c1:83:5d:80:c3:11:31:e3:82:cf:5c:81:2d:a6:39:f0:10:95:6a:e1:c8:17:40:c5:f7
        prune-backups keep-all=1
        username XXX@XXX

esxi: XXX.XXX.local
        server XXX.XXX.local
        username root
        content import
        skip-cert-verification 1

esxi: YYY.XXX.local
        server YYY.XXX.local
        username root
        content import
        skip-cert-verification 1

nfs: DS_migration_row_Proxmox
        export /nfsmigrationproxmox
        path /mnt/pve/DS_migration_row_Proxmox
        server 1.2.2.4
        content images,import
        prune-backups keep-all=1

command i try is this one
Bash:
id=`pvesh get /cluster/nextid`
qm import $id /mnt/pve/DS_migration_row_Proxmox/Clone_XXX/Clone_XXX.vmx --storage R3_Ceph_Datastore --name BOUM
unable to parse volume ID '/mnt/pve/DS_migration_row_Proxmox/Clone_XXX/Clone_XXX.vmx'

Bash:
pvesm list DS_migration_row_Proxmox
Volid Format  Type      Size VMID

as i try to use nfs (for faster copy) and no VMID seen on pvesm , seems that qm import need it

will try qm create and use --import-from for each scsi disk to see if it works better
 
Looking at the man page of the QM:
Code:
qm import <vmid> <source> --storage <string> [OPTIONS]

       Import a foreign virtual guest from a supported import source, such as an ESXi storage.

       <vmid>: <integer> (100 - 999999999)
           The (unique) ID of the VM.

       <source>: <string>
           The import source volume id.

The <source> option must be in the format of <storage_pool>:path



Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox
 
Hey bbgeek17
like i say on first message using qm import with ESXI_FQDN:ha-datacenter/DS_migration_Proxmox/Clone_mavm/Clone_mavm.vmx works and copy all files, but it's slow.
I found the way to import vm skel only (using --scsiXX file=none) using cli and then import disk (qm disk import) from nfs storage (seems faster) but looking at nfs i'm stuck at 40MB/s (v3 or v4.1 , multiple options tested for mount , no change)
I checked and see qemu-img convert in single thread ,
after editing /usr/share/perl5/PVE/QemuServer/QemuImage.pm i added -m 8 -C to options send to qemu-img but no diff on time to convert

BTW i'm stuck here till now

i have 30TB of storage and hundred of vm to move , so trying to find the fastest solution without nfs share i can use to get vm running (nfs from vsan file services is not supported for running vm on it), and like i say vsan storage backed, so no direct path to migrate more smoothly

Indeed was a good try till now, not sure where is bottleneck for importing vmdk to ceph backed storage here
 
Last edited:
for support is it possible to ask some change on importer from esxi from gui to allow a tick button and option to thow for qme-img all caches options available ? i think it will help a lot of people to get descend migration time (on my test on a vm 5H with gui 5 min with a shell script i did). Otherwise i consider my thread ok to solve
 
for support is it possible to ask some change on importer from esxi from gui to allow a tick button and option to thow for qme-img all caches options available ? i think it will help a lot of people to get descend migration time (on my test on a vm 5H with gui 5 min with a shell script i did). Otherwise i consider my thread ok to solve
We are looking at doing the same thing, migrating from vSphere with vSAN to Proxmox. Would you be able to share the shell script? Looking into different migration paths and as of now vSAN makes everything harder...