On April 1, 2021 12:19 pm, Dominic Jäger wrote:
> On Wed, Mar 31, 2021 at 05:12:28PM +0200, Fabian Grünbichler wrote:
>> this is starting to shape up nicely. as promised, I now took a stab at
>> (roughly!) integrating this into our regular flow (see diff below):
>> ....
>>
>> - we could likely drop the separate import_disk API call, and let the
>> `importdisk` CLI command prepare parameters for a regular VM config
>> update
> I think dropping importdisk is a good idea, too.
>
>> I'm sure I've missed some corner cases, as I've only tested create_vm
>> with importing, and not the other newly exposed APIs.
>
> I didn't get the importdisk API to work yet.
> But I'd look closer at stuff like that
> if (!$device_options) {
> $device_options = "$storeid:0"; <=== should be -1
> }
> only if we keep it?
>
> One thing is important: That import parameter was at the place of the default
> storage
yes, I just roughly moved the code to see whether the approach works out
and then tested a single entry point. if there are no objections into
merging it like that, it would definitely need a cleanup and more
testing :)
>
> PVE/API2/Qemu.pm | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/PVE/API2/Qemu.pm b/PVE/API2/Qemu.pm
> index 41e1ab7..73af497 100644
> --- a/PVE/API2/Qemu.pm
> +++ b/PVE/API2/Qemu.pm
> @@ -1469,7 +1469,7 @@ my $update_vm_api = sub {
> PVE::QemuServer::vmconfig_register_unused_drive($storecfg,
> $vmid, $conf, PVE::QemuServer::parse_drive($opt, $conf->{pending}->{$opt}))
> if defined($conf->{pending}->{$opt});
>
> - &$create_disks($rpcenv, $authuser, $conf->{pending}, $arch,
> $storecfg, $vmid, undef, {$opt => $param->{$opt}}, {$opt =>
> $import_devices->{$opt}});
> + &$create_disks($rpcenv, $authuser, $conf->{pending}, $arch,
> $storecfg, $vmid, undef, {$opt => $param->{$opt}}, undef, {$opt =>
> $import_devices->{$opt}});
> } elsif ($opt =~ m/^serial\d+/) {
> if ((!defined($conf->{$opt}) || $conf->{$opt} eq 'socket')
> && $param->{$opt} eq 'socket') {
> $rpcenv->check_vm_perm($authuser, $vmid, undef,
> ['VM.Config.HWType']);
>
>
>> stat shows how much boilerplace/duplication this removes, although there
>> is probably even more potential here since `create_vm` partly duplicates
>> the `update_vm_api` sub that ends up calling `create_disks`:
> Is that urgent?
no, it's just something I noticed since I added the same code in the
same context in two places ;)
_______________________________________________
pve-devel mailing list
pve-devel@lists.proxmox.com
https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel
