In my initial postings, I described a situation in which I had created an
account, but that account was causing the following error to be generated
in /var/adm/messages:
Feb 16 09:09:13 aries nisd[17555]: _svcauth_des: corrupted window from unix.333@mof.nis
Feb 16 09:09:25 aries last message repeated 17 times
Further, the account seemed to have been corrupted somehow; e.g. using what
seemed to be a Bourne shell instead of the configured csh, sparse env,
inability to exec, ls, vi etc. despite having the correct paths to these
executables, didn't recognize directories that did exist etc. All in all,
an account that was very, very sick.
When I had a look at the NIS+ tables, they seemed to be okay on our master
server, but were corrupt on our replica (...strange...). In addition, the
rpc.nisd daemon had died and could not be restarted on the replica, complaining
about corrupt NIS+.
The solution to the problem was to delete the user account in question and
also delete the replica. I then used nismkdir to recreate the replica and
then, once I verified that NIS+ seemed to be working properly again, I created
the user account EXACTLY the way I had created it before. Everything is
working properly now, the user account is uncorrupted, no "corrupted window"
warnings in /var/adm/messages and NIS+ is functioning correctly. Since I
couldn't replicate the problem, it looks like it was a one-off. Don't you
just love those?
I have no idea what happened; whether the account corrupted NIS+ or vice versa
or even whether the two events were related. *big shrug*
Thanks to all who responded. Most pointed out that the user account seemed to
be using the Bourne-shell rather than the configured C-shell, thus the
sparse env and inability to source (which is inbuilt to csh). The solution
may not be the most elegant nor did it tell me what happened, but it worked,
and in my books, that's 99% of the law :)
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:10:16 CDT