[mpich-discuss] MPICH2 on HPC Server 2008 wdir problem

[Addepalli, Srirangam V]

Thank you all. I tried with the -wdir option compute node crashed with blue screen but only after the job started to read/write to network drive. I am assuming it is something to do with MS HPC job scheduler now.
Rangam

Problem signature:
  Problem Event Name:	BlueScreen
  OS Version:	6.0.6001.2.1.0.272.18
  Locale ID:	1033

Additional information about the problem:
  BCCode:	27
  BCP1:	00000000BAAD0075
  BCP2:	FFFFFA6005EF6228
  BCP3:	FFFFFA6005EF5C00
  BCP4:	FFFFFA6004111734
  OS Version:	6_0_6001
  Service Pack:	1_0
  Product:	272_3

Files that help describe the problem:
  C:\Windows\Minidump\Mini051709-04.dmp
  C:\Users\Administrator\AppData\Local\Temp\2\WER-2587978-0.sysdata.xml
  C:\Users\Administrator\AppData\Local\Temp\2\WER78F.tmp.version.txt

Read our privacy statement:
  http://go.microsoft.com/fwlink/?linkid=50163&clcid=0x0409


________________________________________
From: Jayesh Krishna [jayesh at mcs.anl.gov]
Sent: Sunday, May 17, 2009 9:36 AM
To: Addepalli, Srirangam V
Cc: mpich-discuss at mcs.anl.gov
Subject: RE: [mpich-discuss] MPICH2 on HPC Server 2008  wdir problem

 You will also need to set the working directory (-wdir) for your job.

-jayesh

-----Original Message-----
From: mpich-discuss-bounces at mcs.anl.gov [mailto:mpich-discuss-bounces at mcs.anl.gov] On Behalf Of Markus Fischer
Sent: Sunday, May 17, 2009 4:10 AM
To: mpich-discuss at mcs.anl.gov
Subject: Re: [mpich-discuss] MPICH2 on HPC Server 2008 wdir problem

have you tried

  job submit /workdir:\\p\a\t\h

Markus


Addepalli, Srirangam V wrote:
> Hello Jayesh,
>
> I tried both the options suggested.
>
> mpiexec -np 8 -path Z:\BIN\ nwchem.exe siosi3.nw  (Here Z:\BIN is a
> network mapped drive) with MS HPC job scheduler variables
> WorkingDirectory \\Fileshare\BIN\ Using this option i was able to get
> the executable to start but it exits as it is not able find the input
> file.. siosi3.nw
>
> argument  1 = siosi3.nw
>   could not open a file in permanent directory:
>   could not open a file in scratch directory:
>   Both permanent and scratch directory not accessible
>  ******** Fatal Error ********      911
> This option worked till it found the nwchem.exe binary. But executable (nwchem) failed to read the input file from the directory it was supposed to be reading from. I believe these errors are caused by the way the file sharing is setup.
>
> I switched over to the second option and specified the full path
>
> mpiexec -np 8  \\Fileserv\BIN\nwchem.exe siosi3.nw
>
> Working Directory \\Fileserv\BIN\
>
> This kills the compute node ..with no information in the event viewer about what triggered the reboot.same set of jobs work perfectly fine using mpiexec wrapper interface ...
>
> So for i was only able to get it to work using mpiexecwrapper.
>
> mpiexec -np 8 nwchem.exe siosi3.nw  (this works when both nwchem.exe
> and siosi3.nw are located in C:\
> MPICH2\bin)
>
> Rangam
> ________________________________________
> From: Jayesh Krishna [jayesh at mcs.anl.gov]
> Sent: Saturday, May 16, 2009 8:34 PM
> To: Addepalli, Srirangam V
> Cc: mpich-discuss at mcs.anl.gov
> Subject: RE: [mpich-discuss] MPICH2 on HPC Server 2008  wdir problem
>
> Hi,
>
> # Does specifying a complete path to the executable work (mpiexec -n 2 c:\temp\cpi.exe )?
> # Did you try the "-path" option of mpiexec when launching your job (mpiexec -n 2 -path c:\temp cpi.exe) ?
>
> -jayesh
>
> -----Original Message-----
> From: mpich-discuss-bounces at mcs.anl.gov
> [mailto:mpich-discuss-bounces at mcs.anl.gov] On Behalf Of Addepalli,
> Srirangam V
> Sent: Friday, May 15, 2009 5:18 PM
> To: mpich-discuss at mcs.anl.gov
> Subject: [mpich-discuss] MPICH2 on HPC Server 2008 wdir problem
>
> Hello All.
> I am trying to used MPICH2 on windows HPC Cluster.  Using HPC job scheduler i am not able to specify the work directory when i try to submit a job. mpiexec will look at the input files only if preset in  MPICH2/BIN directory. I have tried using the -wdir setting but not able to get mpiexec to work.
> Sincerely
> Rangam
>
>
>
>
> ----------------------------------------------------------------------
> --
>
>
> No virus found in this incoming message.
> Checked by AVG - www.avg.com
> Version: 8.5.325 / Virus Database: 270.12.32/2118 - Release Date:
> 05/16/09 17:05:00
>