[PVE-User] Network device name change (with systemd) under debian after adding/removing line meta: from config file

[PVE-User] Network device name change (with systemd) under debian after adding/removing line meta: from config file

[Laurent CARON]

Hi,

I'm experiencing a rather strange behavior with VM creation on a 7.7 
proxmox cluster.

I do have quite old (migrated from older proxmox versions) VMs as well 
as new VMs (created after the 7.2 version).

Config file:

$ cat /etc/pve/qemu-server/889.conf
boot: order=virtio0;ide2;net0
cores: 1
cpu: host
ide2: none,media=cdrom
machine: q35
memory: 2048
meta: creation-qemu=7.0.0,ctime=1687340877
name: vs-lcaron-testalacon
net0: virtio=C2:B9:B3:FE:6C:B0,bridge=vmbr2000
numa: 1
ostype: l26
scsihw: virtio-scsi-pci
smbios1: uuid=ef269cd3-e7a6-442e-87ad-30b68741003b
sockets: 1
virtio0: VMs_NFS:889/vm-889-disk-0.qcow2,discard=on,size=64G
vmgenid: ac09af11-7012-41eb-99e3-ca1162afc2bd

Using this config, VM installs fine.

Interface name is enp6s18

If I do remove the line 'meta: ...' from the config file, and restart 
the VM's interface is now ens18.

Please note, when using 'meta: ...' lspci | grrp Ethernet is:

06:12.0 Ethernet controller: Red Hat, Inc. Virtio network device

When not using 'meta: ...' lspci is exactly the same.


What is changing behind the scenes when using 'meta: ' in the config file ?


I'm trying to have a consistent naming across VMs, wether the VMs were 
instanciated on an old or a new proxmox version.

Thanks

Re: [PVE-User] Network device name change (with systemd) under debian after adding/removing line meta: from config file

[Fabian Grünbichler]

there was a change in how Qemu sets up PCI slots. we record the version with which the VM was first created exactly to avoid this change during the lifetime of the VM:

https://git.proxmox.com/?p=qemu-server.git;a=commitdiff;h=cc1810363579b8d6bf7dd8d681b831f33209bbb0

> Laurent CARON <lcaron@unix-scripts.info> hat am 21.06.2023 14:22 CEST geschrieben:
> 
>  
> Hi,
> 
> I'm experiencing a rather strange behavior with VM creation on a 7.7 
> proxmox cluster.
> 
> I do have quite old (migrated from older proxmox versions) VMs as well 
> as new VMs (created after the 7.2 version).
> 
> Config file:
> 
> $ cat /etc/pve/qemu-server/889.conf
> boot: order=virtio0;ide2;net0
> cores: 1
> cpu: host
> ide2: none,media=cdrom
> machine: q35
> memory: 2048
> meta: creation-qemu=7.0.0,ctime=1687340877
> name: vs-lcaron-testalacon
> net0: virtio=C2:B9:B3:FE:6C:B0,bridge=vmbr2000
> numa: 1
> ostype: l26
> scsihw: virtio-scsi-pci
> smbios1: uuid=ef269cd3-e7a6-442e-87ad-30b68741003b
> sockets: 1
> virtio0: VMs_NFS:889/vm-889-disk-0.qcow2,discard=on,size=64G
> vmgenid: ac09af11-7012-41eb-99e3-ca1162afc2bd
> 
> Using this config, VM installs fine.
> 
> Interface name is enp6s18
> 
> If I do remove the line 'meta: ...' from the config file, and restart 
> the VM's interface is now ens18.
> 
> Please note, when using 'meta: ...' lspci | grrp Ethernet is:
> 
> 06:12.0 Ethernet controller: Red Hat, Inc. Virtio network device
> 
> When not using 'meta: ...' lspci is exactly the same.
> 
> 
> What is changing behind the scenes when using 'meta: ' in the config file ?
> 
> 
> I'm trying to have a consistent naming across VMs, wether the VMs were 
> instanciated on an old or a new proxmox version.
> 
> Thanks
> 
> _______________________________________________
> pve-user mailing list
> pve-user@lists.proxmox.com
> https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-user

Re: [PVE-User] Network device name change (with systemd) under debian after adding/removing line meta: from config file

[Laurent CARON]

Thanks for the pointer Fabian. This is exactly it.


Laurent

Le 21/06/2023 à 14:35, Fabian Grünbichler a écrit :
> there was a change in how Qemu sets up PCI slots. we record the version with which the VM was first created exactly to avoid this change during the lifetime of the VM:
>
> https://git.proxmox.com/?p=qemu-server.git;a=commitdiff;h=cc1810363579b8d6bf7dd8d681b831f33209bbb0
>
>> Laurent CARON <lcaron@unix-scripts.info> hat am 21.06.2023 14:22 CEST geschrieben:
>>
>>   
>> Hi,
>>
>> I'm experiencing a rather strange behavior with VM creation on a 7.7
>> proxmox cluster.
>>
>> I do have quite old (migrated from older proxmox versions) VMs as well
>> as new VMs (created after the 7.2 version).
>>
>> Config file:
>>
>> $ cat /etc/pve/qemu-server/889.conf
>> boot: order=virtio0;ide2;net0
>> cores: 1
>> cpu: host
>> ide2: none,media=cdrom
>> machine: q35
>> memory: 2048
>> meta: creation-qemu=7.0.0,ctime=1687340877
>> name: vs-lcaron-testalacon
>> net0: virtio=C2:B9:B3:FE:6C:B0,bridge=vmbr2000
>> numa: 1
>> ostype: l26
>> scsihw: virtio-scsi-pci
>> smbios1: uuid=ef269cd3-e7a6-442e-87ad-30b68741003b
>> sockets: 1
>> virtio0: VMs_NFS:889/vm-889-disk-0.qcow2,discard=on,size=64G
>> vmgenid: ac09af11-7012-41eb-99e3-ca1162afc2bd
>>
>> Using this config, VM installs fine.
>>
>> Interface name is enp6s18
>>
>> If I do remove the line 'meta: ...' from the config file, and restart
>> the VM's interface is now ens18.
>>
>> Please note, when using 'meta: ...' lspci | grrp Ethernet is:
>>
>> 06:12.0 Ethernet controller: Red Hat, Inc. Virtio network device
>>
>> When not using 'meta: ...' lspci is exactly the same.
>>
>>
>> What is changing behind the scenes when using 'meta: ' in the config file ?
>>
>>
>> I'm trying to have a consistent naming across VMs, wether the VMs were
>> instanciated on an old or a new proxmox version.
>>
>> Thanks
>>
>> _______________________________________________
>> pve-user mailing list
>> pve-user@lists.proxmox.com
>> https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-user