mic@ut-emx.uucp (Mic Kaczmarczik) (01/17/91)
In article <15631@sdcc6.ucsd.edu> mattson@beowulf.ucsd.edu (Jim Mattson) writes: >Here's the problem. The NeXT is set up as a non-netinfo machine in the NIS >domain 'cse'. NIS binding seems to work just fine. I can ypcat and ypmatch >on the NeXT with no problems. However, things blow up when I make the >following addition to /etc/passwd: > > +@ape-users::0:0::: > >Now, if a user not specified in the local netinfo database tries to log in, >the machine sort of hangs. It hangs in a weird way, though. Telnetting in >will make a connection, but the NeXT won't print a login prompt. I had exactly the same problem when I upgraded our support cube to 2.0 a week ago. NIS worked fine except when there was a +@netgroup in /etc/passwd, at which point all logins hung. Based on some very sketchy Ethernet packet tracing from a nearby Sun, it looked like the machine was generating NIS lookups to its NIS server, but not enough to come even close to enumerating our entire NIS password file (which is pretty huge). I stopped investigating at that point due to time constraints. >If wordse comes to worse, I suppose I could extract all of the users from >the ape-users netgroup and enter them into netinfo with niload. I could >even do this nightly out of cron, so that password changes in the NIS maps >would get propagated to the NeXT. However, this approach seems awfully >kludgy to me. Has anyone else seen this problem? Better yet, has anyone >else solved it? Even though entries with +@netgroup didn't work, entries of the form +username did. Perhaps your cron job could extract all the users from the ape-users group and put a +user line for each one: +user1 +user2 >--jim >-- > Internet: jmattson@ucsd.edu > Bitnet: jmattson@ucsd > UUCP: {ucbvax|rutgers|bellcore}!ucsd!jmattson -- Mic Kaczmarczik | Be wary of strong drink. It can make you Unix/VMS Services | shoot at tax collectors -- and miss. UT Austin Computation Center | remark@{ccwf,emx,bongo} 1-0251 | -- Lazarus Long