Has anybody here used this functionality before? Should I be asking a different list?
On 2011-09-17 13:05, Matthew Cengia wrote:
> Hi all,
>
> I'm using the VFS sbadow_copy2 module, and I think I've hit a bug. This
> config works:
>
> mattcen@adam:SambaTest$ cat works3/smb.conf
> [SambaTest]
> comment = SambaTest
> writable = yes
> path = /srv/SambaTest/share
> #vfs objects = shadow_copy
> vfs objects = shadow_copy2
> shadow:snapdir = /mnt/dates
> shadow:basedir = /srv
> mattcen@adam:SambaTest$ ls /srv
> SambaTest aquota.group aquota.user archive bittorrent cvs lost+found
> lxc nfs share tftp time_machine tmp
> mattcen@adam:SambaTest$ ls /mnt/dates/@GMT-*
> /mnt/dates/@GMT-2011.09.16-22.44.13:
> SambaTest
>
> /mnt/dates/@GMT-2011.09.16-22.44.36:
> SambaTest
>
> /mnt/dates/@GMT-2011.09.16-22.45.24:
> SambaTest
>
> /mnt/dates/@GMT-2011.09.16-23.03.18:
> SambaTest
>
> /mnt/dates/@GMT-2011.09.16-23.03.37:
> SambaTest
>
> /mnt/dates/@GMT-2011.09.16-23.03.56:
> SambaTest
> mattcen@adam:SambaTest$
>
> This config doesn't:
>
> mattcen@adam:SambaTest$ cat broken1/smb.conf
> [SambaTest]
> comment = SambaTest
> writable = yes
> path = /srv/SambaTest/share
> #vfs objects = shadow_copy
> vfs objects = shadow_copy2
> shadow:snapdir = /mnt/dates
> shadow:basedir = /
> mattcen@adam:SambaTest$ ls /
> bin boot cgroup dev etc home initrd.img lib lost+found media mnt
> opt proc root sbin selinux snapshots srv sys tmp usr var vmlinuz
> mattcen@adam:SambaTest$ ls /mnt/dates/@GMT-*
> /mnt/dates/@GMT-2011.09.16-22.44.13:
> srv
>
> /mnt/dates/@GMT-2011.09.16-22.44.36:
> srv
>
> /mnt/dates/@GMT-2011.09.16-22.45.24:
> srv
>
> /mnt/dates/@GMT-2011.09.16-23.03.18:
> srv
>
> /mnt/dates/@GMT-2011.09.16-23.03.37:
> srv
>
> /mnt/dates/@GMT-2011.09.16-23.03.56:
> srv
> mattcen@adam:SambaTest$
>
> The latter config gives me the following in /var/log/samba/log.<client>:
>
> [2011/09/17 00:06:42.354400, 0]
> modules/vfs_shadow_copy2.c:380(convert_shadow2_name)
> convert_shadow2_name: basedir / is not a parent of /srv/SambaTest/share
>
> I'm hitting this code in vfs_shadow_copy2.c, but I don't really
> understand what it's doing:
>
> mattcen@adam:samba$ nl -ba
> samba-3.5.6~dfsg/source3/modules/vfs_shadow_copy2.c | sed -n 376,383p
> 376 /* some sanity checks */
> 377 if (strncmp(basedir, handle->conn->connectpath, baselen) !=
> 0 ||
> 378 (handle->conn->connectpath[baselen] != 0 &&
> handle->conn->connectpath[baselen] != '/')) {
> 379 DEBUG(0,("convert_shadow2_name: basedir %s is not a
> parent of %s\n",
> 380 basedir, handle->conn->connectpath));
> 381 talloc_free(tmp_ctx);
> 382 return NULL;
> 383 }
>
> I'm running Samba on Debian Squeeze:
>
> mattcen@adam:tmp$ lsb_release -idrc
> Distributor ID: Debian
> Description: Debian GNU/Linux 6.0.2 (squeeze)
> Release: 6.0.2
> Codename: squeeze
> mattcen@adam:tmp$ dpkg -l samba | tail -1
> ii samba 2:3.5.6~dfsg-3squeeze5
> SMB/CIFS file, print, and login server for Unix
>
> Checking the Samba Git repo, it looks like the version I'm running is
> from around commit 5529289e1d17d0524dc38aea11e0a3469818753c (at least
> that has the same version of the vfs_shadow_copy2.c file), and I can see
> that the git trunk has a complete rewrite of this module, but I don't
> want to have to recompile my Samba deb after cherry-picking that module.
>
> Can anybody see anything wrong with the way I've configured my smb.conf,
> or does this actually look like a bug (which probably isn't worth
> reporting since this code seems to be almost obsolete)?
>
> --
> Regards,
> Matthew Cengia
--
Regards,
Matthew Cengia
signature.asc
Description: Digital signature
-- To unsubscribe from this list go to the following URL and read the instructions: https://lists.samba.org/mailman/options/samba
