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?
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.
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 fedora-selinux-list@redhat.com http://www.redhat.com/mailman/listinfo/fedora-selinux-list
Or mount -o bind it?
On Saturday 18 December 2004 01:14, Troels Arvin troels@arvin.dk 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.
selinux@lists.fedoraproject.org