[pve-devel] [RFC PATCH 2/4] api: add reboot api call

[Stefan Reiter]

On 8/14/19 4:20 PM, Dominik Csapak wrote:
> this creates a reboot trigger file (inspired by pve-container)
> and relies on the 'qm cleanup' call by the qmeventd to detect
> and restart the vm afterwards
> 
> Signed-off-by: Dominik Csapak <d.csapak at proxmox.com>
> ---
>   PVE/API2/Qemu.pm | 45 +++++++++++++++++++++++++++++++++++++++++++++
>   1 file changed, 45 insertions(+)
> 
> diff --git a/PVE/API2/Qemu.pm b/PVE/API2/Qemu.pm
> index 8be0b7b..d59e23c 100644
> --- a/PVE/API2/Qemu.pm
> +++ b/PVE/API2/Qemu.pm
> @@ -2339,6 +2339,51 @@ __PACKAGE__->register_method({
>   	}
>       }});
>   
> +__PACKAGE__->register_method({
> +    name => 'vm_reboot',
> +    path => '{vmid}/status/reboot',
> +    method => 'POST',
> +    protected => 1,
> +    proxyto => 'node',
> +    description => "Reboot the VM by shutting it down, and starting it again.",
> +    permissions => {
> +	check => ['perm', '/vms/{vmid}', [ 'VM.PowerMgmt' ]],
> +    },
> +    parameters => {
> +	additionalProperties => 0,
> +	properties => {
> +	    node => get_standard_option('pve-node'),
> +	    vmid => get_standard_option('pve-vmid',
> +					{ completion => \&PVE::QemuServer::complete_vmid_running }),
> +	    skiplock => get_standard_option('skiplock'),
> +	    timeout => {
> +		description => "Wait maximal timeout seconds for the shutdown.",
> +		type => 'integer',
> +		minimum => 0,
> +		optional => 1,
> +	    },
> +	    forceStop => {
> +		description => "Make sure the VM stops.",
> +		type => 'boolean',
> +		optional => 1,
> +		default => 0,
> +	    },
> +	},
> +    },
> +    returns => {
> +	type => 'string',
> +    },
> +    code => sub {
> +	my ($param) = @_;
> +
> +	# mark the vm for rebooting, will be caught by qm cleanup
> +	my $vmid = $param->{vmid};
> +	open(my $fh, '>', "/var/lib/qemu-server/$vmid.reboot")
> +	    or die "failed to create reboot trigger file: $!\n";
> +	close($fh);
> +	return PVE::API2::Qemu->vm_shutdown($param);

Testing this, if a backup job (for example) is holding a VM lock, and 
you reboot, the <vmid>.reboot file will be created, but the shutdown 
fails - next "poweroff" will then be a restart.

In general, if the shutdown fails, the file will be left behind, which 
seems like a general issue with the file based approach (in line with 
your concerns on the series' cover).

> +    }});
> +
>   __PACKAGE__->register_method({
>       name => 'vm_suspend',
>       path => '{vmid}/status/suspend',
>