The ABINIT Users Mailing List ( CLOSED )

["Anglade Pierre-Matthieu" <anglade@gmail.com>]

Do you have a nfs on your cluster? From the abinip error message we
may guess that one of the "abinip" was not able to read a file.  Is it
possible that this is the problem?

On 6/25/06, chinafuentes2000@yahoo.com <chinafuentes2000@yahoo.com> wrote:
> Hi
>
> We have problems running abinip in our cluster ( one server and 2 nodes).
> When we try to submit a parallel job, the mpirun give us the following msg:
>
> forrtl: severe (24): end-of-file during read, unit 5, file stdin
> Image              PC        Routine            Line        Source
> abinip             08F13C2C  Unknown               Unknown  Unknown
> abinip             08F13724  Unknown               Unknown  Unknown
> abinip             08EE0AED  Unknown               Unknown  Unknown
> abinip             08EACC40  Unknown               Unknown  Unknown
> abinip             08EAD0E3  Unknown               Unknown  Unknown
> abinip             08EC6A30  Unknown               Unknown  Unknown
> abinip             081FFD76  Unknown               Unknown  Unknown
> abinip             080483A8  Unknown               Unknown  Unknown
> abinip             08048268  Unknown               Unknown  Unknown
> abinip             08F1A32C  Unknown               Unknown  Unknown
> abinip             08048121  Unknown               Unknown  Unknown
> -----------------------------------------------------------------------------
> It seems that [at least] one of the processes that was started with
> mpirun did not invoke MPI_INIT before quitting (it is possible that
> more than one process did not invoke MPI_INIT -- mpirun was only
> notified of the first one, which was on node n0).
>
> mpirun can *only* be used with MPI programs (i.e., programs that
> invoke MPI_INIT and MPI_FINALIZE).  You can use the "lamexec" program
> to run non-MPI programs over the lambooted nodes.
> -----------------------------------------------------------------------------
> forrtl: error (78): process killed (SIGTERM)
>
>
> our cluster was mounted using Oscar package.
>
> we submit the job in the following way:
>
> mpirun -np 2 /abinitbin-4.6.5/abinip < tparal_1.files > logp
>
> here is the lamnodes command output:
>
> n0      lqc.uach.mx:1:origin,this_node
> n1      lqc01.fcq.uach.mx:1:
> n2      lqc02.fcq.uach.mx:1:
>
> we downloaded and installed the following version:
> pclinux_ifc-4.6.5.tar.gz
>
> we use the binaries included in the .tgz. ( we did not compile)
>
> Why abinip does not run in the nodes ?
>
> Is it necessary to compile abinit or just use the binaries of the package ?
>
>
> the following runs OK:
>
> mpirun -np 1 /abinitbin-4.6.5/abinip < tparal_1.files > logp
> mpirun n0 /abinitbin-4.6.5/abinip < tparal_1.files > logp
>
> the following does not runs:
>
> mpirun N /abinitbin-4.6.5/abinip < tparal_1.files > logp
> mpirun n1 /abinitbin-4.6.5/abinip < tparal_1.files > logp
>
> Thanks for your help.


--
Pierre-Matthieu Anglade