[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]


Miquel van Smoorenburg writes:

 > >I suspect it has something to do with the way shadow passwords work in
 > >Solaris and in Linux. However, on an adjacent Slackware box it works
 > >flawlessly. What can be the problem?
 > Is the output of "ypcat passwd" the same? Does the Slackware box
 > run libc5, libc6, what version etc?

It's a libc5 Sackware box (I'm not sure what Slackware version, but I
checked that it's libc5). Yes - the output of "ypcat passwd" there is
the same.

Now I'm sure that the problem is with shadow - I ltraced login, and
saw that it crypts the password I enter with the "##" salt, and then
compares it to the "##username" string it got from getpwent() as the
password. Therefore, it doesn't even look for a shadow password. The
machine uses shadow passwords for the root & local accounts. I thought 
that maybe it would look for a shadow password if the password was "x" 
and not the "##username" thing, so I changed the last line in
/etc/passwd to "+:x:::::", but <sigh> as always that didn't help.

 > >I tried to upgrade to the latest nis distribution from slink (and it
 > >took the libc along with it) but it did not help.
 > Can you do "ypcat passwd.adjunct" (as root!) ? Perhaps you need to
 > fiddle with /etc/nsswitch.conf

No. (No such map...) But I also can't do it from any other box that
already is using NIS successfully, such as our SunOS and Solaris
boxes! I checked and the map _is there_, and there's indeed no
shadow.byname map. AAACK!

Alex Shnitman
alexsh@linux.org.il   UIN 188956
http://alexsh.home.ml.org  --  PGP key here
http://www.debian.org  --  and the OS here

Reply to: