> We have a heterogenous network with different machines running
>OS4.1.1 and OS4.1.2. (We want to upgrade to one version, and this
>inquiry is to clarify some things which have confused us.) We are not
>running NIS but using a nameserver and modified libc.so library. The
>most pressing problem is that after putting 4.1.2 on a Sparc 1
>which had been running 4.1.1, we find that the command
>
>cd /net/machine
>
>hangs for one specific machine (actually the nameserver and
>itself running 4.1.2), but
>for no others. The command hangs, and the Sparc 1 also hangs in that
>logins hang after the motd. The OS was installed by copying from
>another machine and making appropriate modifications. We also had
>a similar phenomenon on an IPX, but that was fixed by trying another
>kernel, or so we hope. Kernel change doesn't help with the
>Sparc 1, although from a software point of view, it should be
>essentially the same as the IPX.
>
The cause of this problem was a bit subtle but not very difficult
to find once we looked a little more closely. Because of the
way we had installed the OS, the root partition had `frozen'
mount points. These looked like bonafide automount points but
in fact prevented atuomounting from working properly. We discovered
this by disabling automounting and checking for mount points and links.
When these were removed, automounting worked properly.
This sort of problem can arise if you dump your file
systems without shutting down and then have to restore them after a
crash.
>There is a related phenomenon which is also puzzling. On some
>machines running OS4.1.2, using
>
>cd /net/machine
>
>to a machine running OS4.1.1 works, and
>
>ls
>
>then works, but also yields the message
>
>ld.so: warning: usr/lib/libc.so.1.6.1 has older revision than expected 7
>
>However, this does not occur in all circumstances for 4.1.2 machines.
>It also does not occur if
>one is in a subdirectory of /usr/net/machine which the base machine
>has permission to mount.
>
>Can anyone shed any light? Needless to say we have thoroughly
>examined most of the obvious things like export tables, etc.
This was simpler still. My path was different on different machines.
The message appeared when the system first checked the current
directory. In some
cases, the current directory was last in my path, so it
was never searched. I am still not quite sure
why the library from the mounted machine came into it, but that
is not entirely unreasonable.
Leonard Evens len@math.nwu.edu 708-491-5537
Dept. of Mathematics, Northwestern Univ., Evanston, IL 60208
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:08:07 CDT