phil@eecs.nwu.edu (William LeFebvre) (10/06/90)
I just made the mistake of activating YP on our multimax running UMAX 4.3. The multimax is client only, being bound (right now) to a Sun 3/80 master running 4.0.3. Before doing this drastic deed, I made sure that all the stuff that was in /etc/{rpc,services,protocols} was really in the YP maps. I did NOT add any '+' lines to passwd, preferring to leave all the account explicitly specified. Remember: this system worked fine before I ran ypbind. Now that YP is running, csh exhibits the following bizarre behavior: % cd % ls | cat > ~luser/a cat: read error: Bad file number % (where the current user is in fact "luser". The following does what you would expect: % cd % ls | cat > ~/a % tcsh 5.12 is showing similar weirdness when it comes to username expansion. I tried recompiling tcsh and the freshly compiled version has the same problems. Any clues? I am also very annoyed that Encore's "rpcinfo" does not have a "-d" option. Right now, the only way to turn off YP is to reboot. If I had a "-d" I could tell the portmapper to delete the YP mappings, but I don't so I'm pretty much stuck. Grrr....... William LeFebvre Computing Facilities Manager and Analyst Department of Electrical Engineering and Computer Science Northwestern University <phil@eecs.nwu.edu>
george@CIS.OHIO-STATE.EDU (George M. Jones) (10/11/90)
From: phil@eecs.nwu.edu (William LeFebvre) Newsgroups: comp.sys.encore Date: 5 Oct 90 20:58:07 GMT Reply-To: phil@eecs.nwu.edu (William LeFebvre) Organization: EECS Department, Northwestern University Now that YP is running, csh exhibits the following bizarre behavior: % cd % ls | cat > ~luser/a cat: read error: Bad file number % (where the current user is in fact "luser". The following does what you would expect: % cd % ls | cat > ~/a % Any clues? William LeFebvre Computing Facilities Manager and Analyst Department of Electrical Engineering and Computer Science Northwestern University <phil@eecs.nwu.edu> I'm a little slow in catching up... I suggest you try a few experiments with getpwnam. I suspect that what is happening is something like the [t]csh calling getpwnam, which happily goes out and does a yp call, and low and behold there is no user named "luser" in the yp password file, so the routine happily reports a failure to the shell, but you were able to log in which means that login(1) probabibly uses an older/ypless version of getpwnam, so it succeeds in reading from /etc/passwd. BTW, if anybody has a compiled and working version of a recent tcsh for Umax 4.3 with YP I'd like to hear about it/get your diffs/ftp a copy. ---George Jones
rackow@ANTARES.MCS.ANL.GOV (10/11/90)
The problem happens with Encore's supplied csh as well as the tcsh that Bill (or someone) ported. I have the same problem on our Multimax at Argonne. I have had yp running for some time, but never tested that situation. Why can't encore get it right? --Gene