[gmx-users] GROMACS on glacier.westgrid.ca

[Mark Abraham]

Paymon Pirzadeh wrote:
> Hello all,
> A follow up from my previous problems on running the gromacs on glacier
> turned out that I had problems in installation. The guide in the manual
> was not clear enough. So, I installed the code fine this time, so, I get
> errors from 4.0.4 version. 
> I ran the code interactively on the master node and everything went
> smooth and fine without any problems. But as soon as I started running
> it in parallel through que line, job gets killed. Following are the
> messages I received:
> 
> PBS Job Id: 4686006.teva.westgrid.ubc
>>>> Job Name:   sixsite-test
>>>> Exec host:  ice1_2/1+ice1_2/0+ice1_1/1+ice1_1/0
>>>> Aborted by PBS Server
>>>> Job cannot be executed
> 
> The error messages I receive in error file:
>>>> Killed by signal 2.^M
>>>> Killed by signal 2.^M
>>>> Killed by signal 2.^M
> 
> Running on 4 processors.
>>>> Starting run at: Mon May 25 20:29:34 PDT 2009
>>>> p2_11876:  p4_error: Timeout in establishing connection to remote
>>>> process: 0
>>>> p0_4273:  p4_error: net_recv read:  probable EOF on socket: 1
>>>> Job finished at: Mon May 25 20:34:40 PDT 2009
> 
> All the clues show that the program had started working, initial
> conditions were reviewed by the code, but suddenly everything crashed. I

So what was at the end of stderr and stdout from the job? There's no 
GROMACS output given here, so we've got no evidence of a non-cluster 
GROMACS problem.

> really do not know where the error originates from!!! This is while my
> friend has compiled the code in the same way as I did, and she is
> running fine on another cluster.

Make sure you've followed the instructions for parallel compilation and 
execution for your facility and for GROMACS. Make a shell script that 
does the build so that you can do a build reproducibly, check it has the 
same problems and then ask for advice on it from your sysadmins or from 
this list.

Mark