[pve-devel] [PATCH v4 qemu-server 1/12] migration: only migrate disks used by the guest

[Fiona Ebner]

Am 16.06.23 um 11:56 schrieb Aaron Lauterer:
> When scanning all configured storages for disk images belonging to the
> VM, the migration could easily fail if a storage is not available, but
> enabled. That storage might not even be used by the VM at all.
> 
> By not scanning all storages and only looking at the disk images
> referenced in the VM config, we can avoid unnecessary failures.
> Some information that used to be provided by the storage scanning needs
> to be fetched explicilty (size, format).
> 
> Behaviorally the biggest change is that unreferenced disk images will
> not be migrated anymore. Only images in the config or in a snapshot will
> be migrated.
> 
> The tests have been adapted accordingly.
> 
> Signed-off-by: Aaron Lauterer <a.lauterer at proxmox.com>
> ---
> changes since v3: now it only removes the storage scanning

Ideally, this patch should come after recognizing pending disks in
foreach_volid(), because like this, the patch breaks picking up the
pending disks during migration until the later patch is also there. But
if it's too much work to reorder, feel free to leave it.

> 
>  PVE/QemuMigrate.pm                    | 49 ++++-----------------------
>  test/MigrationTest/QemuMigrateMock.pm | 10 ++++++
>  test/run_qemu_migrate_tests.pl        | 11 +++---
>  3 files changed, 22 insertions(+), 48 deletions(-)
> 
> diff --git a/PVE/QemuMigrate.pm b/PVE/QemuMigrate.pm
> index 09cc1d8..5f4f402 100644
> --- a/PVE/QemuMigrate.pm
> +++ b/PVE/QemuMigrate.pm
> @@ -312,49 +312,6 @@ sub scan_local_volumes {
>  	    $abort = 1;
>  	};
>  
> -	my @sids = PVE::Storage::storage_ids($storecfg);
> -	foreach my $storeid (@sids) {
> -	    my $scfg = PVE::Storage::storage_config($storecfg, $storeid);
> -	    next if $scfg->{shared} && !$self->{opts}->{remote};
> -	    next if !PVE::Storage::storage_check_enabled($storecfg, $storeid, undef, 1);
> -
> -	    # get list from PVE::Storage (for unused volumes)
> -	    my $dl = PVE::Storage::vdisk_list($storecfg, $storeid, $vmid, undef, 'images');
> -
> -	    next if @{$dl->{$storeid}} == 0;
> -
> -	    my $targetsid = PVE::JSONSchema::map_id($self->{opts}->{storagemap}, $storeid);
> -	    if (!$self->{opts}->{remote}) {
> -		# check if storage is available on target node
> -		my $target_scfg = PVE::Storage::storage_check_enabled(
> -		    $storecfg,
> -		    $targetsid,
> -		    $self->{node},
> -		);
> -
> -		die "content type 'images' is not available on storage '$targetsid'\n"
> -		    if !$target_scfg->{content}->{images};
> -
> -	    }
> -
> -	    my $bwlimit = $self->get_bwlimit($storeid, $targetsid);
> -
> -	    PVE::Storage::foreach_volid($dl, sub {
> -		my ($volid, $sid, $volinfo) = @_;
> -
> -		$local_volumes->{$volid}->{ref} = 'storage';
> -		$local_volumes->{$volid}->{size} = $volinfo->{size};
> -		$local_volumes->{$volid}->{targetsid} = $targetsid;
> -		$local_volumes->{$volid}->{bwlimit} = $bwlimit;
> -
> -		# If with_snapshots is not set for storage migrate, it tries to use
> -		# a raw+size stream, but on-the-fly conversion from qcow2 to raw+size
> -		# back to qcow2 is currently not possible.
> -		$local_volumes->{$volid}->{snapshots} = ($volinfo->{format} =~ /^(?:qcow2|vmdk)$/);

Sorry, only noticed now, but this one is dropped, but it shouldn't be.
See 5eca0c36 ("sync_disks: Always set 'snapshots' for qcow2 and vmdk
volumes") for the rationale. You can just re-do it further down when you
have the format and please also copy the comment.

> -		$local_volumes->{$volid}->{format} = $volinfo->{format};
> -	    });
> -	}
> -
>  	my $replicatable_volumes = !$self->{replication_jobcfg} ? {}
>  	    : PVE::QemuConfig->get_replicatable_volumes($storecfg, $vmid, $conf, 0, 1);
>  	foreach my $volid (keys %{$replicatable_volumes}) {
> @@ -407,6 +364,12 @@ sub scan_local_volumes {
>  	    $local_volumes->{$volid}->{ref} = 'storage' if $attr->{is_unused};
>  	    $local_volumes->{$volid}->{ref} = 'generated' if $attr->{is_tpmstate};
>  
> +	    $local_volumes->{$volid}->{bwlimit} = $self->get_bwlimit($sid, $targetsid);
> +	    $local_volumes->{$volid}->{targetsid} = $targetsid;
> +
> +	    ($local_volumes->{$volid}->{size}, $local_volumes->{$volid}->{format})
> +		= PVE::Storage::volume_size_info($storecfg, $volid);

If you want it to fit on one line:
$local_volumes->{$volid}->@{qw(size format)} =
PVE::Storage::volume_size_info($storecfg, $volid);

> +
>  	    $local_volumes->{$volid}->{is_vmstate} = $attr->{is_vmstate} ? 1 : 0;
>  
>  	    $local_volumes->{$volid}->{drivename} = $attr->{drivename}
> diff --git a/test/MigrationTest/QemuMigrateMock.pm b/test/MigrationTest/QemuMigrateMock.pm
> index 94fe686..9034d10 100644
> --- a/test/MigrationTest/QemuMigrateMock.pm
> +++ b/test/MigrationTest/QemuMigrateMock.pm
> @@ -240,6 +240,16 @@ $MigrationTest::Shared::storage_module->mock(
>  
>  	delete $source_volids->{$volid};
>      },
> +    volume_size_info => sub {
> +	my ($scfg, $volid) = @_;
> +	my ($storeid, $volname) = PVE::Storage::parse_volume_id($volid);
> +
> +	for my $v ($source_vdisks->{$storeid}->@*) {
> +	    return wantarray ? ($v->{size}, $v->{format}, $v->{used}, $v->{parent}) : $v

Above line should end with ": $v->{size}"?

> +		if $v->{volid} eq $volid;
> +	}
> +	die "could not find '$volid' in mock 'source_vdisks'\n";
> +    },
>  );
>  
>  $MigrationTest::Shared::tools_module->mock(
> diff --git a/test/run_qemu_migrate_tests.pl b/test/run_qemu_migrate_tests.pl
> index 090449f..7a9d7ea 100755
> --- a/test/run_qemu_migrate_tests.pl
> +++ b/test/run_qemu_migrate_tests.pl
> @@ -708,7 +708,6 @@ my $tests = [
>  	},
>      },
>      {
> -	# FIXME: Maybe add orphaned drives as unused?
>  	name => '149_running_orphaned_disk_targetstorage_zfs',
>  	target => 'pve1',
>  	vmid => 149,
> @@ -729,10 +728,11 @@ my $tests = [
>  	},
>  	expected_calls => $default_expected_calls_online,
>  	expected => {
> -	    source_volids => {},
> +	    source_volids => {
> +		'local-dir:149/vm-149-disk-0.qcow2' => 1,
> +	    },
>  	    target_volids => {
>  		'local-zfs:vm-149-disk-10' => 1,
> -		'local-zfs:vm-149-disk-0' => 1,
>  	    },
>  	    vm_config => get_patched_config(149, {
>  		scsi0 => 'local-zfs:vm-149-disk-10,format=raw,size=4G',
> @@ -765,10 +765,11 @@ my $tests = [

There's another FIXME above here, that should be removed

>  	},
>  	expected_calls => $default_expected_calls_online,
>  	expected => {
> -	    source_volids => {},
> +	    source_volids => {
> +		'local-dir:149/vm-149-disk-0.qcow2' => 1,
> +	    },
>  	    target_volids => {
>  		'local-lvm:vm-149-disk-10' => 1,
> -		'local-dir:149/vm-149-disk-0.qcow2' => 1,
>  	    },
>  	    vm_config => get_patched_config(149, {
>  		scsi0 => 'local-lvm:vm-149-disk-10,format=raw,size=4G',