Re: Compilation on an Opteron Server SuSE Linux Enterprise System Version 8

From: Marek Dynowski (marek.dynowski_at_student.uni-tuebingen.de)
Date: Thu Oct 21 2004 - 03:03:54 CDT

Hi Gengbin,
now I tried what you said and it seems that it is a problem with the cluster.

1. the command ./namd2 namdconfig (../NAMD_2.5_Linux-i686-TCP/namd2
dyn0000-tcl.namd) produces the message:
 
Charm++: standalone mode (not using charmrun)
Segmentation fault

If I use charmrun with the following command:

2. ../NAMD_2.5_Linux-i686-TCP/charmrun +p1 ../NAMD_2.5_Linux-i686-TCP/namd2
dyn0000-tcl.namd

This error message occurs:

Charmrun: error on request socket--
Socket closed before recv.

But if I use charmrun with the option ++local

3. ./NAMD_2.5_Linux-i686-TCP/charmrun +p1 ../NAMD_2.5_Linux-i686-TCP/namd2
dyn0000-tcl.namd ++local

It works but only at the main node so I think that there is a problem with the
setup of the cluster but why doesn't it work if I use only namd without
charm?

Thanks, Marek

On Wednesday 20 October 2004 17:24, Gengbin Zheng wrote:
> Hi Marek,
>
> first try to run it sequentially before you try something else, just
> run it without charmrun:
> ./namd2 namdconfig
> In most case this will tells you what error you get (like missing shared
> library due to incompatable kernel library, etc)
> If this works, run it parallel but only on your local desktop, like:
> ./charmrun +p2 ./namd2 namdconfig ++local
>
> If this still run, it is likely the problem of your cluster setup that
> prevent you from running charmrun across different machines.
>
> Gengbin
>
> Marek Dynowski wrote:
> >On Wednesday 20 October 2004 15:27, you wrote:
> >> I'm running namd in a opteron cluster with fedora 2.0 x86_64. I was
> >>also not able to compile it due to similar problems, but the binary
> >>
> >> NAMD_2.5_Linux-i686-TCP
> >>
> >>which is available in the namd site runs well. Of course, it runs as a
> >>32bit application.
> >> Leandro.
> >
> >Hi Leandro / all,
> >
> >thanks for the quick answer. I tried the binary NAMD_2.5_Linux-i686-TCP,
> > but now I get the following error message:
> >
> >
> >Charmrun: error on request socket--
> >Socket closed before recv.
> >
> >I found the same error message in the Charm++ FAQ:
> >
> >Charmrun: error on request socket--
> >Socket closed before recv.
> >This means that the node program died without informing charmrun about it,
> >which typically means a segmentation fault while in the interrupt handler
> > or other critical communications code. This indicates severe corruption
> > in Charm++'s data structures, which is likely the result of a heap
> > corruption bug in your program. Re-linking with "-memory paranoid" may
> > clarify the true problem.
> >
> >So I tried to compile NAMD/Charm to solve the error above ?
> >Do you have any suggestions ?

-- 
------------------------
Marek Dynowski
ZMBP, Plant Physiology
University of Tuebingen
Auf der Morgenstelle 1
D-72076 Tübingen
Ph: +49 7071 29 74643
------------------------

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:37:55 CST