[pve-devel] [PATCH v3 storage 2/4] storage_migrate: pull out snapshot decision

[Fabian Ebner]

Am 22.12.21 um 14:52 schrieb Fabian Grünbichler:
> into new top-level helper for re-use with remote migration.
> 
> Signed-off-by: Fabian Grünbichler <f.gruenbichler at proxmox.com>
> ---
> new in v3
> 
>   PVE/Storage.pm | 14 ++++++++++----
>   1 file changed, 10 insertions(+), 4 deletions(-)
> 
> diff --git a/PVE/Storage.pm b/PVE/Storage.pm
> index 05be3dd..4f3a44a 100755
> --- a/PVE/Storage.pm
> +++ b/PVE/Storage.pm
> @@ -643,6 +643,14 @@ my $volname_for_storage = sub {
>       }
>   };
>   
> +# whether a migration snapshot is needed for a given storage
> +sub storage_migrate_snapshot {
> +    my ($cfg, $storeid) = @_;
> +    my $scfg = storage_config($cfg, $storeid);
> +
> +    return $scfg->{type} eq 'zfspool' || $scfg->{type} eq 'btrfs';
> +}
> +
>   sub storage_migrate {
>       my ($cfg, $volid, $target_sshinfo, $target_storeid, $opts, $logfunc) = @_;
>   
> @@ -688,10 +696,8 @@ sub storage_migrate {
>   
>       my $migration_snapshot;
>       if (!defined($snapshot)) {
> -	if ($scfg->{type} eq 'zfspool' || $scfg->{type} eq 'btrfs') {
> -	    $migration_snapshot = 1;
> -	    $snapshot = '__migration__';
> -	}
> +	$migration_snapshot = storage_migrate_snapshot->($cfg, $target_storeid);

The call is wrong, because of the ->

And why base the decision on the target storage rather than keeping it 
based on the source storage? That breaks migration from e.g. lvm-thin to 
btrfs, because it now tries to use a snapshot and doesn't find any 
common transfer format anymore.

> +	$snapshot = '__migration__' if $migration_snapshot;
>       }
>   
>       my @formats = volume_transfer_formats($cfg, $volid, $target_volid, $snapshot, $base_snapshot, $with_snapshots);