On Sat, 2013-06-01 at 11:43 +0200, Lukas Slebodnik wrote:
> On (31/05/13 21:29), Jakub Hrozek wrote:
> >On Wed, 2013-05-29 at 16:09 +0200, Lukas Slebodnik wrote:
> >> On (29/05/13 08:44), Simo Sorce wrote:
> >> >On Wed, 2013-05-29 at 11:30 +0200, Lukas Slebodnik wrote:
> >> >> On (29/05/13 11:07), Lukas Slebodnik wrote:
> >> >> >ehlo,
> >> >> >
> >> >> >Function krb5_cc_get_full_name is called only as a way to
validate that,
> >> >> >we have the right cache. Instead of returned name, location
will be returned
> >> >> >from function cc_dir_cache_for_princ.
> >> >> >
> >> >> >https://fedorahosted.org/sssd/ticket/1936
> >> >> >
> >> >> >Patch is attached.
> >> >> >
> >> >> >LS
> >> >>
> >> >> self NACK
> >> >>
> >> >> this patch store to cache DIR:/run/user/325600000/krb5cc
> >> >> ^^^^
> >> >> missing colon?
> >> >
> >> >No, this is the correct form.
> >> I found out, that it is a correct form.
> >> Problem was with checking ccname in function sss_krb5_cc_file_path.
> >>
> >> New patches attached.
> >>
> >> LS
> >
> >Sorry for the reply from gmail. My OTP token decided the best password
> >for me on a Friday evening is "Err", so I can't access my
redhat.com
> >account at the moment.
> >
> >These patches break one assumption we want to keep -- if there is a user
> >logged in and the same user logs in for example from another terminal,
> >they should have the same ccache. With your patches, I'm getting a new
> >one when I log in simultaneously.
> >
> >I haven't tested that, but I guess this is because path to collection is
> >always passed to the krb5_child now. I think that in the case user is
> >already logged in (in krb5 code we denote this with "ccache is
active"),
> >then you should pass the full path to the ccache to the krb5_child.
> >
> Simo wrote in ticket comment (
https://fedorahosted.org/sssd/ticket/1936#comment:10)
>
> > Do we really want to store only DIR:/run/user/$uid/krb5cc/ to cache?
> Yes, this is exactly what we want as a ccache.
>
> LS
Yes, I saw that comment and I agree with Simo.
But I think we should examine the ccache collection and in the case
there already is a valid cache present there, we should reuse it just
like we did before.
btw this actually sounds like something the krb5 libs should be doing on
their own, I can't think of a situation where you'd want two ccaches for
exactly the same principals in the same collection..