[pve-devel] [PATCH proxmox-perl-rs v2 02/11] add Perl mapping for OCI container image parser/extractor

[Wolfgang Bumiller]

On Wed, Jun 25, 2025 at 09:59:08AM +0200, Filip Schauer wrote:
> On 24/06/2025 14:51, Wolfgang Bumiller wrote:
> > > +    #[export]
> > > +    pub fn parse_and_extract_image(
> > > +        oci_tar_path: &str,
> > > +        rootfs_path: &str,
> > > +    ) -> Result<Option<Config>, Error> {
> > > +        match proxmox_oci::parse_and_extract_image(oci_tar_path, rootfs_path) {
> > > +            Ok(config) => Ok(Some(config.unwrap_or_default())),
> > > +            Err(err) => match err {
> > > +                ProxmoxOciError::ParseError(ParseError::NotAnOciImage(_)) => Ok(None),
> > ^ Why are we doing this?
> 
> In the Perl code of the create_vm API method at PVE::API2::LXC, if a
> .tar file is used as the container template,
> PVE::RS::OCI::parse_and_extract_image is first attempted.
> It returns undef, when the .tar file is not an OCI image. Thus we deduce
> that the file is an LXC template. If some other kind of error occurrs,
> (e.g. I/O error) then the Perl code should die.
> 
> The point is, there needs to be a way to differentiate between the file
> actually being an LXC template, and some general failure.

Given how the errors in the oci-spec work I think it would make more
sense to do this detection separately by simply looking for an
`oci-layout` file in the archive (or looking for all of `oci-layout`,
`blobs`, `index.json` at only the top level).

An an error in the OCI data should not cause us to try to extract this
as a regular container either.

In the long run, this needs to be a separate content type anyway, since
don't want to just extract tar files like this. Rather, we want to later
also add the ability to use the base image via overlayfs.