yum/rpm failing with pre or post script errors + sshd

seth vidal skvidal at fedoraproject.org
Thu Apr 10 03:42:06 UTC 2008


On Thu, 2008-04-10 at 07:14 +0800, John Summerfield wrote:
> seth vidal wrote:
> > On Wed, 2008-04-09 at 19:07 +1000, David Timms wrote:
> >> Rex Dieter wrote:
> >>> David Timms wrote:
> >>>
> >>>> I'm having trouble on only one of two test machines.
> >>>>
> >>>> when I yum update, I see a lot of:
> >>>>    Updating : at-spi
> >>>> error: %post(at-spi...) scriptlet failed, exit status 255.
> >>> Bugzilla please (against at-spi).  pkg scriplet's should never fail.
> >> Actually every package generates these same errors. I don't think we 
> >> really want 500 bugzilla items. I don't really want to update at the 
> >> moment because it just causes more failed rpm transactions.
> >>
> >> Any ideas for what the underlying cause might be ?
> >>
> >> Actually, attempting to start sshd on that machine also fails:
> >> # service sshd start
> >> Starting sshd: /usr/sbin/sshd: error while loading shared libraries: 
> >> libwrap.so.0: cannot open shared object file: No such file or directory
> >>    [FAILED]
> >>
> >> # sshd
> >> sshd re-exec requires execution with an absolute path
> >>
> >> Think it is related ?
> > 
> > just for fun, put selinux in permissive mode, then retry it.
> > 
> > I wonder if something is unhappy in one of your protections.
> 
> Also,
>   1. get one of the updates in place with rpm --noscripts.
>   2. Find the relevent scripts with rpm -q scripts
>   3. Run the relevant bit without >/dev/null
> 
> Red Hat thinks we don't need error messages.
> 


"Red Hat" has nothing to do with it. Killing the error message is normal
for things which can't error out in any useful way.

The errors in this case wouldn't show much, the logs - specifically
selinux logs, would.

-sv





More information about the test mailing list