Parallel Processing Problems Fluent 13

lstebe · February 1, 2012, 11:33

Hello guys,

i´m workin at Technische Universität Darmstadt.
Everytime the people from our chair want to calculate larger cases we have to buy processing time at our supercomputer.
Well, in our students-workroom we have 10pc´s between 4 and 8 cores per engine so i would like to build-up an hpc cluster in here with Win server 2008 hpc. So far no problem.
I installed one as a headnode and one as a compute node.
Fluent on the headnode and the fluent installation-path shared over enterprise network (boss will give me the money for private network when i succeed with the trials), also shared a working-path.
added the nodes to the cluster can see them in heat map, can run diagnosis functions without failure but won´t get connections from HN to CN over fluent.

Of course i added the network paths to the fluent-launcher, chose microsoft job scheduler, ethernet and msmpi.

when i choose 4 parallel processes fluent starts correctly on the local machine.
if i choose 6 or 8 cores i get an

"Job has been submitted. ID: 359.
Waiting for CCP scheduler@headnode to start msmpi nodes ...
Job 359 is Running."

Fluent hangs and no reaction. can you help me to find out what fault i coud have overlooked?

thank you for your answers!
best regards lindsay

lstebe · February 14, 2012, 12:58

anybody to help me.
i would appreciate answers or questions to solve it.
thanks a lot

best regards

Zigainer · February 15, 2012, 04:15

How many nodes does your local machine have?
Are the other nodes in ~/.ssh/know_hosts on your local machine?

lstebe · February 15, 2012, 10:54

hey,
yes nodes are known
i tried to start a job from the headnote, from a workstation (win7) thats connected to the cluster that runs fluent on its own and from the (one trial) compute node that runs win server 08 r2 hpc, besides all pc run 64- bit systems so that the batch mode should not be necessary as i ´ve read in the anises "help-me"
HN 4 cores
WS 4 cores
CN 4 cores

thank you

February 1, 2012, 11:33	Parallel Processing Problems Fluent 13	#1
lstebe New Member Join Date: Feb 2012 Posts: 3 Rep Power: 14	Hello guys, i´m workin at Technische Universität Darmstadt. Everytime the people from our chair want to calculate larger cases we have to buy processing time at our supercomputer. Well, in our students-workroom we have 10pc´s between 4 and 8 cores per engine so i would like to build-up an hpc cluster in here with Win server 2008 hpc. So far no problem. I installed one as a headnode and one as a compute node. Fluent on the headnode and the fluent installation-path shared over enterprise network (boss will give me the money for private network when i succeed with the trials), also shared a working-path. added the nodes to the cluster can see them in heat map, can run diagnosis functions without failure but won´t get connections from HN to CN over fluent. Of course i added the network paths to the fluent-launcher, chose microsoft job scheduler, ethernet and msmpi. when i choose 4 parallel processes fluent starts correctly on the local machine. if i choose 6 or 8 cores i get an "Job has been submitted. ID: 359. Waiting for CCP scheduler@headnode to start msmpi nodes ... Job 359 is Running." Fluent hangs and no reaction. can you help me to find out what fault i coud have overlooked? thank you for your answers! best regards lindsay

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
FLUENT Parallel Post Processing	john_cse	FLUENT	0	October 20, 2011 06:49
Parallel running of Fluent	Bhanu Gupta	FLUENT	3	April 7, 2011 10:32
bubbly flow and parallel processing	mvee	FLUENT	0	September 12, 2007 06:08
Fluent parallel	poovanna	FLUENT	0	March 24, 2007 00:22
error parallel fluent session	Diet	FLUENT	2	January 27, 2005 13:31

February 14, 2012, 12:58		#2
lstebe New Member Join Date: Feb 2012 Posts: 3 Rep Power: 14	anybody to help me. i would appreciate answers or questions to solve it. thanks a lot best regards

February 15, 2012, 04:15		#3
Zigainer Senior Member Join Date: May 2011 Location: Germany Posts: 130 Rep Power: 15	How many nodes does your local machine have? Are the other nodes in ~/.ssh/know_hosts on your local machine?

February 15, 2012, 10:54		#4
lstebe New Member Join Date: Feb 2012 Posts: 3 Rep Power: 14	hey, yes nodes are known i tried to start a job from the headnote, from a workstation (win7) thats connected to the cluster that runs fluent on its own and from the (one trial) compute node that runs win server 08 r2 hpc, besides all pc run 64- bit systems so that the batch mode should not be necessary as i ´ve read in the anises "help-me" HN 4 cores WS 4 cores CN 4 cores thank you