AMBER Archive (2002)

Subject: Re: ES40 hangs amber7 jobs.

From: Sanjeev B.S. (sanjeev_at_mbu.iisc.ernet.in)
Date: Mon Sep 02 2002 - 21:29:45 CDT

Hello,
        Thanks to Mr. Chuck Schneider's help and advice, the problem has
been apparently fixed by installing the latest MPI available at:
http://www.compaq.com/hpc/software/soft_download_regCMPI.html
The job used to hand in a couple of minutes, and now it has been running
for 15hours. Thanks again to Scheider and Cladwell for the help and
advice.

Sincerely,
-Sanjeev

On Fri, 30 Aug 2002, James W. Caldwell wrote:

We have had problems similar to what you describe on our system of
ES40s (Tru 64 5.1 unpatched) occasionally.

The most likely culprit in your system is NFS problems, the best way
to test this is to run each Amber job completely on a machines's local
disk system then move the files to the user's directory after it finishes.
Once we started doing that our problem with "orphan" jobs cleared up.

best,
jim

On Fri, 30 Aug 2002, Sanjeev B.S. wrote:

>Hello,
> We are trying to use ES40s for amber7. When I run a simulation,
>after a short while it hangs. Processes are running for enternity, and no
>o/p is written whatsoever. No other processes are running that could be
>hindering. OS was freashly installed and latest patches were applied. I
>face similar problems with IBM SP3 also, but I am not very sure of the
>reason as someother jobs which require lots of space and memory run
>togther. I am baffled with this ES40 behavior. We have access to ES40s
>only for certain amount of time, so I would be grateful if anyone can
>suggest me a way out.
>Thanks very much,
>-Sanjeev
>