[pve-devel] [PATCH storage 3/9] plugin: dir: handle ova files for import

[Dominik Csapak]

On 4/17/24 12:52, Fiona Ebner wrote:
> Am 16.04.24 um 15:18 schrieb Dominik Csapak:
>> since we want to handle ova files (which are only ovf+vmdks bundled in a
>> tar file) for import, add code that handles that.
>>
>> we introduce a valid volname for files contained in ovas like this:
>>
>>   storage:import/archive.ova/disk-1.vmdk
>>
>> by basically treating the last part of the path as the name for the
>> contained disk we want.
>>
>> we then provide 3 functions to use for that:
>>
>> * copy_needs_extraction: determines from the given volid (like above) if
>>    that needs extraction to copy it, currently only 'import' vtype +
>>    defined format returns true here (if we have more options in the
>>    future, we can of course easily extend that)
>>
>> * extract_disk_from_import_file: this actually extracts the file from
>>    the archive. Currently only ova is supported, so the extraction with
>>    'tar' is hardcoded, but again we can easily extend/modify that should
>>    we need to.
>>
>>    we currently extract into the import storage in a directory named:
>>    `.tmp_<pid>_<targetvmid>` which should not clash with concurrent
>>    operations (though we do extract it multiple times then)
>>
> 
> Could we do "extract upon upload", "tar upon download" instead? Sure
> some people surely want to drop the ova manually, but we could tell them
> they need to extract it first too. Depending on the amount of headache
> this would save us, it might be worth it.

we could, but this opens a whole other can of worms, namely
what to do with conflicting filenames for different ovas?

we'd then either have to magically match the paths from the ovfs
to some subdir that don't overlap

or we'd have to abort everytime we encounter identical disk names

IMHO this would be less practical than just extract on demand...

> 
>>    alternatively we could implement either a 'tmpstorage' parameter,
>>    or use e.g. '/var/tmp/' or similar, but re-using the current storage
>>    seemed ok.
>>
>> * cleanup_extracted_image: intended to cleanup the extracted images from
>>    above, including the surrounding temporary directory
>>
>> we have to modify the `parse_ovf` a bit to handle the missing disk
>> images, and we parse the size out of the ovf part (since this is
>> informal only, it should be no problem if we cannot parse it sometimes)
>>
>> Signed-off-by: Dominik Csapak <d.csapak at proxmox.com>
>> ---
>>   src/PVE/API2/Storage/Status.pm |  1 +
>>   src/PVE/Storage.pm             | 59 ++++++++++++++++++++++++++++++++++
>>   src/PVE/Storage/DirPlugin.pm   | 13 +++++++-
>>   src/PVE/Storage/OVF.pm         | 53 ++++++++++++++++++++++++++----
>>   src/PVE/Storage/Plugin.pm      |  5 +++
>>   5 files changed, 123 insertions(+), 8 deletions(-)
>>
>> diff --git a/src/PVE/API2/Storage/Status.pm b/src/PVE/API2/Storage/Status.pm
>> index f7e324f..77ed57c 100644
>> --- a/src/PVE/API2/Storage/Status.pm
>> +++ b/src/PVE/API2/Storage/Status.pm
>> @@ -749,6 +749,7 @@ __PACKAGE__->register_method({
>>   				'efi-state-lost',
>>   				'guest-is-running',
>>   				'nvme-unsupported',
>> +				'ova-needs-extracting',
>>   				'ovmf-with-lsi-unsupported',
>>   				'serial-port-socket-only',
>>   			    ],
>> diff --git a/src/PVE/Storage.pm b/src/PVE/Storage.pm
>> index f8ea93d..bc073ef 100755
>> --- a/src/PVE/Storage.pm
>> +++ b/src/PVE/Storage.pm
>> @@ -2189,4 +2189,63 @@ sub get_import_metadata {
>>       return $plugin->get_import_metadata($scfg, $volname, $storeid);
>>   }
>>   
> 
> Shouldn't the following three functions call into plugin methods
> instead? That'd seem much more future-proof to me.

could be, i just did not want to extend the plugin api for that
but as fabian wrote, maybe we should put them in qemu-server
altogether for now?

(after thinking about it a bit, i'd be in favor of putting it in
qemu-server, because mainly i don't want to add to the plugin api further)

what do you think @fiona @fabian?

> 
>> +sub copy_needs_extraction {
>> +    my ($volid) = @_;
>> +    my ($storeid, $volname) = parse_volume_id($volid);
>> +    my $cfg = config();
>> +    my $scfg = storage_config($cfg, $storeid);
>> +    my $plugin = PVE::Storage::Plugin->lookup($scfg->{type});
>> +
>> +    my ($vtype, $name, $vmid, $basename, $basevmid, $isBase, $file_format) =
>> +	$plugin->parse_volname($volname);
>> +
>> +    return $vtype eq 'import' && defined($file_format);
> 
> E.g this seems rather hacky, and puts a weird coupling on a future
> import plugin's parse_volname() function (presence of $file_format).

would it be better to check the volid again for '.ova/something$' ?
or do you have a better idea?
(especially if we want to have this maybe in qemu-server)

> 
>> +}
>> +
>> +sub extract_disk_from_import_file {
>> +    my ($volid, $vmid) = @_;
>> +
>> +    my ($storeid, $volname) = parse_volume_id($volid);
>> +    my $cfg = config();
>> +    my $scfg = storage_config($cfg, $storeid);
>> +    my $plugin = PVE::Storage::Plugin->lookup($scfg->{type});
>> +
>> +    my ($vtype, $name, undef, undef, undef, undef, $file_format) =
>> +	$plugin->parse_volname($volname);
>> +
>> +    die "only files with content type 'import' can be extracted\n"
>> +	if $vtype ne 'import' || !defined($file_format);
>> +
>> +    # extract the inner file from the name
>> +    if ($volid =~ m!${name}/([^/]+)$!) {
>> +	$name = $1;
>> +    }
>> +
>> +    my ($source_file) = $plugin->path($scfg, $volname, $storeid);
>> +
>> +    my $destdir = $plugin->get_subdir($scfg, 'import');
>> +    my $pid = $$;
>> +    $destdir .= "/.tmp_${pid}_${vmid}";
>> +    mkdir $destdir;
>> +
>> +    ($source_file) = $source_file =~ m|^(/.*)|; # untaint
>> +
>> +    run_command(['tar', '-x', '-C', $destdir, '-f', $source_file, $name]);
>> +
>> +    return "$destdir/$name";
>> +}
>> +
>> +sub cleanup_extracted_image {
>> +    my ($source) = @_;
>> +
>> +    if ($source =~ m|^(/.+/\.tmp_[0-9]+_[0-9]+)/[^/]+$|) {
>> +	my $tmpdir = $1;
>> +
>> +	unlink $source or $! == ENOENT or die "removing image $source failed: $!\n";
>> +	rmdir $tmpdir or $! == ENOENT or die "removing tmpdir $tmpdir failed: $!\n";
>> +    } else {
>> +	die "invalid extraced image path '$source'\n";
>> +    }
>> +}
>> +
>>   1;
>> diff --git a/src/PVE/Storage/DirPlugin.pm b/src/PVE/Storage/DirPlugin.pm
>> index 4dc7708..50ceab7 100644
>> --- a/src/PVE/Storage/DirPlugin.pm
>> +++ b/src/PVE/Storage/DirPlugin.pm
>> @@ -260,14 +260,25 @@ sub get_import_metadata {
>>       # NOTE: all types must be added to the return schema of the import-metadata API endpoint
>>       my $warnings = [];
>>   
>> +    my $isOva = 0;
>> +    if ($path =~ m!\.ova!) {
> 
> Would be nicer if parse_volname() would return the $file_format and we
> chould check for that. Also missing the $ in the regex, so you'd
> mismatch a weird filename like ABC.ovaXYZ.ovf or?

yeah the $ is missing, and yes, we could return ova/ovf as format there
as we want to change the 'needs extracting' check anyway


> 
>> +	$isOva = 1;
>> +	push @$warnings, { type => 'ova-needs-extracting' };
>> +    }
>>       my $res = PVE::Storage::OVF::parse_ovf($path, $isOva);
>>       my $disks = {};
>>       for my $disk ($res->{disks}->@*) {
>>   	my $id = $disk->{disk_address};
>>   	my $size = $disk->{virtual_size};
>>   	my $path = $disk->{relative_path};
>> +	my $volid;
>> +	if ($isOva) {
>> +	    $volid = "$storeid:$volname/$path";
>> +	} else {
>> +	    $volid = "$storeid:import/$path",
>> +	}
>>   	$disks->{$id} = {
>> -	    volid => "$storeid:import/$path",
>> +	    volid => $volid,
>>   	    defined($size) ? (size => $size) : (),
>>   	};
>>       }
>> diff --git a/src/PVE/Storage/OVF.pm b/src/PVE/Storage/OVF.pm
>> index 4a322b9..fb850a8 100644
>> --- a/src/PVE/Storage/OVF.pm
>> +++ b/src/PVE/Storage/OVF.pm
>> @@ -85,11 +85,37 @@ sub id_to_pve {
>>       }
>>   }
>>   
>> +# technically defined in DSP0004 (https://www.dmtf.org/dsp/DSP0004) as an ABNF
>> +# but realistically this always takes the form of 'bytes * base^exponent'
> 
> The comment wrongly says 'bytes' instead of 'byte' (your test examples
> confirm this).
> 
>> +sub try_parse_capacity_unit {
>> +    my ($unit_text) = @_;
>> +
>> +    if ($unit_text =~ m/^\s*byte\s*\*\s*([0-9]+)\s*\^\s*([0-9]+)\s*$/) {
> 
> Fun regex :P
> 
>> +	my $base = $1;
>> +	my $exp = $2;
>> +	return $base ** $exp;
>> +    }
>> +
>> +    return undef;
>> +}
>> +
> 
> (...)
> 
>> diff --git a/src/PVE/Storage/Plugin.pm b/src/PVE/Storage/Plugin.pm
>> index deaf8b2..ea069ab 100644
>> --- a/src/PVE/Storage/Plugin.pm
>> +++ b/src/PVE/Storage/Plugin.pm
>> @@ -654,6 +654,11 @@ sub parse_volname {
>>   	return ('backup', $fn);
>>       } elsif ($volname =~ m!^snippets/([^/]+)$!) {
>>   	return ('snippets', $1);
>> +    } elsif ($volname =~ m!^import/([^/]+\.ova)\/([^/]+)$!) {
>> +	my $archive = $1;
>> +	my $file = $2;
>> +	my (undef, $format, undef) = parse_name_dir($file);
>> +	return ('import', $archive, 0, undef, undef, undef, $format);
> 
> So we return the same $name for different things here? Not super happy
> with that either. If we were to get creative we could say the archive is
> the "base" of the image, but surely also comes with caveats.

i'll change this in a v2 should not be necessary

> 
>>       } elsif ($volname =~ m!^import/([^/]+$PVE::Storage::IMPORT_EXT_RE_1)$!) {
>>   	return ('import', $1);
>>       } elsif ($volname =~ m!^import/([^/]+\.(raw|vmdk|qcow2))$!) {