Can you suggest a way to get SELinux back in order? [resolved]

Doug H. fedoraproject.org at wombatz.com
Sun Jan 17 00:35:48 UTC 2016


On Fri, 2016-01-15 at 08:08 -0800, Doug H. wrote:
> After about a day of not having crontabs running I realized that
> SELinux was stopping both user and root crontab jobs.  I messed with
> it
> and then discovered:
> 
> https://bugzilla.redhat.com/show_bug.cgi?id=1298192
> (Recent dnf updates caused crontabs to be restricted.  Reboot to last
> kernel fixes that.)
> 
> I added myself to that bug but I had already broken SELinux to the
> point where it would not work properly even if I rebooted to the
> previous kernel.  I was willing to be stupid/brave about messing with
> it since I figure that if I can't get it back in order then I don't
> know enough about it.  A full relabel takes about ten minutes on my
> system, so it is not terrible.
> 
> I have been with "Fedora" since it was "Redhat 5.0" but only enabled
> SELinux after upgrading to F23.  This bug keeps making me want to go
> back to "disabled" but then I come back to it and do some more
> googling.
> 
> My last attempt to fix it was with:
> 
> (reformatted to easier reading)
> dnf reinstall 
> libselinux-utils 
> selinux-policy 
> libselinux-devel 
> libselinux-python3 
> libselinux.i686 
> libselinux-python 
> selinux-policy-targeted 
> rpm-plugin-selinux 
> libselinux.x86_64
> 
> Then a reboot to do the relabel and run with "permissive" to debug.
> It
> had lots of issues.  I suspect that most were simply that I was doing
> working as the admin for my machine and that was not properly set.
> 
> My question:
> 
> Can anybody suggest a series of steps to put my F23 SELinux
> installation in line with the default workstation install of F23?
> 
> Note that I was previously running:
> setsebool -P httpd_enable_homedirs 1
> setsebool -P httpd_read_user_content 1
> I happen to have a personal use web server running out of /home/httpd
> and I would rather leave it there since /home is a partition.  The
> point being that I am fine with running the basic allows that SELinux
> Troubleshooter identified when I first enabled it.

Ok, seems that `dnf reinstall` might have been the issue.

I am back up with SELinux enforcing after:

dnf remove selinux-policy selinux-policy-targeted
rm -rf /etc/selinux
touch /.autorelabel

I first booted to the recent/current kernel and all was good with
crontabs still blocked as expected so I rebooted to 4.2.8-300 and now
all is working and SELinux is back in the picture.  Will now wait for
the bug to be fixed.


-- 
Doug H.


More information about the users mailing list