[pve-devel] [PATCH container] use monitor commands to freeze on pure-v2 setups

[Stoiko Ivanov]

Gave the patch a spin on a v2-only system.
* it fixes the issue as described

we had a short chat off-list with Wolfgang - and considered restructuring
the code a bit (currently the patch makes parts of PVE::CGroup dead code)

but for alleviating the immediate issue:
Tested-By: Stoiko Ivanov <s.ivanov at proxmox.com>
Reviewed-By: Stoiko Ivanov <s.ivanov at proxmox.com>


On Mon, 25 Jan 2021 12:05:32 +0100
Wolfgang Bumiller <w.bumiller at proxmox.com> wrote:

> even an explicit query for the 'unified' cgroup hierarchy
> does not work on lxc when run in a pure v2 environment via
> the kernel options
> 
>     systemd.unified_cgroup_hierarchy=1 cgroup_no_v1=all
> 
> so let's use the freeze commands provided by lxc we already
> have code for...
> 
> Signed-off-by: Wolfgang Bumiller <w.bumiller at proxmox.com>
> See-also: https://forum.proxmox.com/threads/short-read-on-command-socket-error-when-running-backup-of-lxc-containers.82978/
> ---
>  src/PVE/LXC.pm | 12 ++++++++++--
>  1 file changed, 10 insertions(+), 2 deletions(-)
> 
> diff --git a/src/PVE/LXC.pm b/src/PVE/LXC.pm
> index e7a6951..7e84a0e 100644
> --- a/src/PVE/LXC.pm
> +++ b/src/PVE/LXC.pm
> @@ -2463,12 +2463,20 @@ sub get_lxc_version() {
>  
>  sub freeze($) {
>      my ($vmid) = @_;
> -    PVE::LXC::CGroup->new($vmid)->freeze_thaw(1);
> +    if (PVE::CGroup::cgroup_mode() == 2) {
> +	PVE::LXC::Command::freeze($vmid, 30);
> +    } else {
> +	PVE::LXC::CGroup->new($vmid)->freeze_thaw(1);
> +    }
>  }
>  
>  sub thaw($) {
>      my ($vmid) = @_;
> -    PVE::LXC::CGroup->new($vmid)->freeze_thaw(0);
> +    if (PVE::CGroup::cgroup_mode() == 2) {
> +	PVE::LXC::Command::unfreeze($vmid, 30);
> +    } else {
> +	PVE::LXC::CGroup->new($vmid)->freeze_thaw(0);
> +    }
>  }
>  
>  1;