Does anyone have experience with getting Generic NQS 3.50 running on
Dec Alpha under DU3.2? I tried to upgrade a very old version of
Monsanto NQS we were running. Because of conflicts I had to delete
the old version and if I can not get the Generic NQS running I will have
to reinstall the old version.
The process went as follows:-
It did not recognise the machine ID but osf1 worked fine.
The compile gave many warning about define "HAS_64BIT_LONG" in
autoconf.h 17. Should I worry about this and fix it.
When I got to "qmgr start nqs", it failed with not recognising the
machine. It had not made nmap in SETUP. I ran nmapmgr as I did
long ago with the old version. It then loaded the daemons.
When I try to run it even with the simple
qsub
date
^D
it gives empty STDIN.o* and STDIN.e* files and mails root that it
aborted "Aborting signal is : 11" or similar (I did not write the exact
message down but the number was 11). I mailed the nqs-support list 2
days ago and the only replies suggested it was something to do with
finding stty or similar in .profile, .login etc. This rang bells
as I had some similar problem with the old version. I had similarly
installed Generic NQS on a IBM RS6K where it works fine, so I played there
to clean up all my .profiles etc to get clean runs with no terminal messages
and with ksh, csh and sh (I have not done bash yet, although SETUP selected
bash to use). I then used this experience to play on the alpha. I
have tried all sorts of ideas - using -x or not, various shells, putting
#!/bin/sh or whatever at the top of a script to run as "qsub script".
I varied the .profile, .login, .cshrc, .kshrc and checked them very carefully.
Result - no change in behaviour.
I found just one change on the RS6K. If you use -x, ENVIRONMENT is not
set to BATCH. Without -x it is. I do not think this was the behaviour
with the old Monsanto NQS. I use this variable to switch out stty etc
calls in .profile etc.
Does anyone have a clue? I'll summarise.
Cheers, Brian.
--
Associate Professor Brian Salter-Duke (Brian Duke)
School of Mathematical and Physical Sciences, Northern Territory University,
Darwin, NT 0909, Australia. Phone 089-466702
e-mail: b_duke_at_lacebark.ntu.edu.au or b_duke_at_quoll.ntu.edu.au
Received on Sun Jun 30 1996 - 03:18:14 NZST