Subject: Bug? execvp w/dead NFS server (fwd)
To: None <current-users@NetBSD.ORG>
From: Brian C. Grayson <firstname.lastname@example.org>
Date: 09/29/1997 23:13:42
This languished on tech-userlevel for an entire week without a
peep. Anyone on current-users know if this is a bug or a
feature? Note that, IIRC, many things, like shells, use execvp,
so the behavior mentioned below can be quite confusing --
``/usr/local/bin/acrobat exists, and I can cksum it, so why does
the shell complain that a resource is temporarily unavailable when
I try to run it without specifying a full path name, while it
works if a full path name is specified?''
I can provide a transcript if the situation is unclear.
----- Forwarded message from Brian C. Grayson -----
Subject: Bug? execvp w/dead NFS server
Date: Mon, 22 Sep 1997 08:47:53 -0500 (CDT)
I don't know if this is a bug or not, but it's definitely
confusing behavior, at least to me:
One NFS server is down. I happen to have directories from
this server near the front of my PATH. When this server is
down, any execvp()s of programs that are from later components
of the path fail, because execvp() is bailing out on the EAGAIN
(Resource temporarily unavailable) return code from its execve
attempts using the unavailable directory name, rather than
checking the remaining components of the path.
I could change my path in all running programs, but in some
that's not possible without killing the program, which is not
always an option.
So, bug or feature? :)
(I'm running a July 28 userland, and a Sept 17 kernel, on
-i386. Yes, it's time to upgrade userland.)
----- End of forwarded message from Brian C. Grayson -----
Brian Grayson (email@example.com)
Graduate Student, Electrical and Computer Engineering
The University of Texas at Austin
Office: ENS 406 (512) 471-8011
Finger firstname.lastname@example.org for PGP key.