Quoting Laine Stump (laine@xxxxxxxxx): > On 10/05/2011 06:33 AM, Daniel P. Berrange wrote: > >On Tue, Oct 04, 2011 at 12:49:03PM -0500, Serge E. Hallyn wrote: > >>Quoting Serge E. Hallyn (serge.hallyn@xxxxxxxxxxxxx): > >>>isos are read-only, so libvirt doesn't need to chown them. In one of > >>>our testing setups, libvirt uses mirrorred isos. Since libvirt chowns > >>>the files, (and especially does not chown them back) the mirror refuses > >>>to update the iso. > >>> > >>>This patch prevents libvirt from chowning files. > >>> > >>>Does this seem reasonable? > >>any feedback on this? Does it seem ok? > >Unfortunately while this would fix the use case you describe, it would > >also break other use cases. > > > >What we really need todo with the DAC driver is replace all the > >chown() code, with code that sets ACLs instead. Well actually we > >would need to keep the chown code as a fallback for filesystems > >which don't support ACLs, but as long as we prefer ACLs by default > >that'd be OK. > > > >Of course when we have ACLs, we'd only need to grant 'r' to the > >file for CDROMs which would be better than what we do now. > > > In the meantime, I think you can solve the problem with your mirror > by mounting the share read-only. When the filesystem is read-only, > libvirt will attempt the chown/chgrp and fail, but notice the > failure is due to a r/o (or root-squash) filesystem, and ignore the > failure. Thanks for the suggestion, Laine. I suspect the dir they're using needs to be writeable, but I'll ask if it is feasible to use a read-only bind mount of the directory. thanks, -serge -- libvir-list mailing list libvir-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/libvir-list