as part of my ongoing investigation on sssd-1.8.93's bugs, here is
another distilled item:
- when /var/lib/sss/db is missing, sssd does not start. That's ok,
though the syslog error message "Cannot load configuration database"
is *very* sparse. (It could, for example, say what file it failed to
open, and strerror(errno)).
- when /var/lib/sss/mc is missing, sssd does start, and emits a
warning message into sssd_nss.log, but only at debug_level=9.
This needs to be "louder". (Perhaps terminate sssd, like in the /db
- Can't sssd just create all the directories it needs in /var/lib/sss
- if /var/lib/sss/mc is missing, the inability to read /mc/group
goes in silence. (Well, that's perhaps a "bug" of id(1)
and getent(1).) It requires that one checks the return code
and errno to see that getgrid returns EBADMSG.
krb5_child was crashing on my 32bit system when debugging was enabled
(and to make it more interesting, only when compiled with -O2). We were
using %llu to print uid_t values, which didn't match with the real size
of uid_t on 32bit systems. This patch adds an explicit cast to unsigned
The name context was not being initialized for local provider
domains because it was handled after skipping over the back-end
initialization routine. This patch moves the name context init
routine to occur earlier.
this patch added the checks requested in ticket #1382 to the PAC
responder. The check itself can be found in the commom responder code.
It can be used by all responder, but currently only the PAC responder
I took a quite strict default here, i.e. only root is allowed to access
the PAC responder by default. Is this too restrictive?