[mpich-discuss] Problem using Mpich2 on Vista machines

Larry Adams ladams at platform.com
Wed Jun 24 16:02:29 CDT 2009


Jack,
 
Are you using any scheduler around this?  There are workarounds in some
Windows schedulers, if this becomes a problem.  If it's a small doit
yourself install, you may be ok by simply using the passphase and
properly configuring the smpd service.
 
I must also add that DNS (forward and reverse) lookup is essential.
Make sure you have each of your subnet's properly configured by your DNS
Server for both forward and reverse lookups.
 
Another good troubleshooting step is to do the following:
 
telnet myotherhost smpd_port
 
You should receive a connection
 
mpiexec -hosts 1 myotherhost cmd /c hostname
 
Regards,
 
Larry Adams 
Senior Systems Engineer 
Platform Computing 
Tele: (586) 510-0007 
Cell: (586) 899-1138 
Skype: TheWitness 
 

  _____  

From: mpich-discuss-bounces at mcs.anl.gov
[mailto:mpich-discuss-bounces at mcs.anl.gov] On Behalf Of Fairchild, Jack
Sent: Wednesday, June 24, 2009 4:18 PM
To: mpich-discuss at mcs.anl.gov
Subject: Re: [mpich-discuss] Problem using Mpich2 on Vista machines



-status does show SMPD running.

The ipaddress produces the same error.

Is there a way to debug the error further?

 

Jack Fairchild, CFPS  |  Senior Fire Protection Designer  |  Ballinger
|  v 215.446.0596 |   f  215.446.0597  |  833 Chestnut Street  Suite
1400    Philadelphia, PA   19107  |  www.ballinger-ae.com

 

From: mpich-discuss-bounces at mcs.anl.gov
[mailto:mpich-discuss-bounces at mcs.anl.gov] On Behalf Of Jayesh Krishna
Sent: Wednesday, June 24, 2009 4:01 PM
To: Fairchild, Jack
Cc: mpich-discuss at mcs.anl.gov
Subject: Re: [mpich-discuss] Problem using Mpich2 on Vista machines

 

Hi,

 

# Is SMPD installed on all the machines (Does "smpd -status" show that
SMPD is running on all the machines)?

# Can you try specifying an ipaddress instead of a hostname while
running your job (mpiexec -n 2 -host IPADDRESS_OF_HOST hostname)?

 

Regards,

Jayesh

 

  _____  

From: mpich-discuss-bounces at mcs.anl.gov
[mailto:mpich-discuss-bounces at mcs.anl.gov] On Behalf Of Fairchild, Jack
Sent: Wednesday, June 24, 2009 2:44 PM
To: mpich-discuss at mcs.anl.gov
Subject: Re: [mpich-discuss] Problem using Mpich2 on Vista machines

Thanks for the quick response Larry.  Windows fire wall is off.  I can
ping the machines, and can access the machines through explorer.  The
pass phrases are all the same "behappy" and since it's all networked, my
login and passwords are the same.  Just in case, I did run the smpd
-register command to no avail.  We're attempting to work around the
problem now by cloning one of the working hard drives.

 

Jack Fairchild

 

From: mpich-discuss-bounces at mcs.anl.gov
[mailto:mpich-discuss-bounces at mcs.anl.gov] On Behalf Of Larry Adams
Sent: Wednesday, June 24, 2009 3:37 PM
To: mpich-discuss at mcs.anl.gov
Subject: Re: [mpich-discuss] Problem using Mpich2 on Vista machines

 

Jack,

 

Try the firewall first, you have to unblock the smpd service port.
Beyond that, you have to verify that smpd service is installed,  and if
it's another host, you will either have to register a pass phrase, or
your account on the execution node to be able to continue.  If your
password ever changes, you have to re-perform this step.

 

Regards,

 

Larry Adams 
Senior Systems Engineer 
Platform Computing 
Tele: (586) 510-0007 
Cell: (586) 899-1138 
Skype: TheWitness 

 

 

  _____  

From: mpich-discuss-bounces at mcs.anl.gov
[mailto:mpich-discuss-bounces at mcs.anl.gov] On Behalf Of Fairchild, Jack
Sent: Wednesday, June 24, 2009 3:26 PM
To: mpich-discuss at mcs.anl.gov
Subject: [mpich-discuss] Problem using Mpich2 on Vista machines

I've debugged as far as I can:

 

Invoking the command: mpiexec -validate -host myhost  returns:

Aborting: connection to smpd rejected

 

Invoking mpiexec -file myfile  returns:

Aborting: unable to connect to myhost

 

Some background.

 

All computers are running Vista Ultimate.  The computers are connected
by a 100mbs LAN.  A few months ago, I set up 4 machines for this use.
They all still connect without issue.

At this point I need to add some more, but am getting these errors.  All
machines have smpd version 1.0.8.  I am an administrator on all
machines, and installed Mpich2 as the administrator.  UAC was turned off
and remains off, all known firewalls are also off.  A path variable to
the bin directory  was added.

 

I ran smpd -d and all of the parameters match those of the working
computers.

 

What am I missing here?  I appreciate your valuable time.

 

Jack Fairchild 

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/mpich-discuss/attachments/20090624/328a02bf/attachment-0001.htm>


More information about the mpich-discuss mailing list