FC3 PostgreSQL update

[Troels Arvin]

Hello,

I installed the updated PostgreSQL for FC3 today. Before the update,
there were no problems, but now - when I try to start PostgreSQL - it
fails, and I get the following line in my /var/log/messages:

kernel: audit(1103292268.189:0): avc:  denied  { search } for  pid=13607
exe=/usr/bin/postgres name=mnt dev=hda5 ino=179521
scontext=root:system_r:postgresql_t tcontext=system_u:object_r:mnt_t
tclass=dir

Due to a disk-space problem, my /var/lib/pgsql is a symlink to
/mnt/hda1/pgsql

My /mnt/hda1 is unlabeled, and I guess this should be changed, but to
what? - Should my /mnt/hda1 be labeled system_u:object_r:root_t?

How come a PostgreSQL update breaks what used to work?

[Chuck Anderson]

On Fri, Dec 17, 2004 at 03:14:52PM +0100, Troels Arvin wrote:
> Due to a disk-space problem, my /var/lib/pgsql is a symlink to
> /mnt/hda1/pgsql

Can you mount hda1 directly on /var/lib/pgsql instead of using a 
symlink?

[Troels Arvin]

On Fri, 17 Dec 2004 09:44:44 -0500, Charles R. Anderson wrote:
> > Due to a disk-space problem, my /var/lib/pgsql is a symlink to
> > /mnt/hda1/pgsql
>
> Can you mount hda1 directly on /var/lib/pgsql instead of using a 
> symlink?

No, my /mnt/hda1 holds other directories which aren't PostgreSQL related.

[Daniel J Walsh]

Charles R. Anderson wrote:
> On Fri, Dec 17, 2004 at 03:14:52PM +0100, Troels Arvin wrote:
> > Due to a disk-space problem, my /var/lib/pgsql is a symlink to
> > /mnt/hda1/pgsql
>
> Can you mount hda1 directly on /var/lib/pgsql instead of using a 
> symlink?
>
> --
> fedora-selinux-list mailing list
> [e-mail]
> http://www.redhat.com/mailman/listinfo/fedora-selinux-list

Or mount -o bind it?

[Russell Coker]

On Saturday 18 December 2004 01:14, Troels Arvin  wrote:
> Due to a disk-space problem, my /var/lib/pgsql is a symlink to
> /mnt/hda1/pgsql

Sym-links interfere with SE Linux labelling.

If you were to run "restorecon -R /var/lib/pgsql/something" then it should 
work.

But the best thing to do is to just not do that.  This is one of the reasons 
why I often promove LVM.  LVM allows you to more effectively manage your 
storage without those sym-link hacks that cause so much trouble with SE Linux 
(and lots of other things too).  This isn't going to fix your current 
problem, but may help you next time.  LVM is really good anyway and is 
something you will benefit from a lot even if you don't use SE Linux.
> How come a PostgreSQL update breaks what used to work?

I don't know how it could have worked previously without PostgreSQL having 
search access to mnt_t.