[OpenAFS-devel] Re: on sockets and devices in AFS

[Derrick Brashear]

things like unionfs mask files; the world seems to have not ended yet. I'm n=
ot worried

Derrick


On Sep 16, 2010, at 7:21 PM, Andrew Deason <adeason@sinenomine.net> wrote:

> On Thu, 16 Sep 2010 11:56:02 -0400
> chas williams - CONTRACTOR <chas@cmf.nrl.navy.mil> wrote:
>=20
>> on old clients, the file would show up with the prefix visible so you
>> wouldnt blindly just write into that file via some program/script.
>> yes, a user could see it but they hopefully wonder about it before
>> deciding to randomly file some empty file with a strange prefix they
>> found (until you update the fileservers).
>=20
> Maybe make the file a symlink to something (invalid) to actively prevent
> this? Could make it a mountpoint to an invalid volume name, maybe.
>=20
> I'm not sure how doing this solves derrick's point "2", anyway. If you
> make the special file with some prefix, the name the application gives
> still masks any "real" file in AFS that has the same name. That is,
> someone creates a socket 'foo', I think you were proposing we create a
> file on the server called .__afs_socket_foo ? We still mask 'foo',
> though.
>=20
> If you just create a symlink with the actual name and make it point to
> something non-dereferencable, though, maybe...
>=20
> I don't know; I don't like masking data, but maybe it wouldn't really be
> such a problem.
>=20
> --=20
> Andrew Deason
> adeason@sinenomine.net
>=20
> _______________________________________________
> OpenAFS-devel mailing list
> OpenAFS-devel@openafs.org
> https://lists.openafs.org/mailman/listinfo/openafs-devel