Learning Forum

[Alan]

Dear all,I hope you are having a nice weekend!nI was trying to submit a Fluent job in a LSF server by using:n fluent 3d -g -t4 -i ###.jou -scheduler_tight_couplingnThe job started successfully but stops very soon becaus of:n Host key verification failed. Error: It seems ssh is trying to verify authenticity of c38b06. Please resolve it and try again! Client interrupted.nI checked online, some were using export SSH_SPAWN=0 before the fluent submission command to avoid SSH verificaiton. But, I came across another problem:n mpirun:rsh:Command not foundnAny inputs and suggestions are greatly appreciated!nthanks in advance!nAlann

[Rob]

Sounds like permissions on the cluster: ie the host can't see/connect to the nodes. n

[Alan]

Sounds like permissions on the cluster: ie the host can't see/connect to the nodes.https://forum.ansys.com/discussion/comment/97449#Comment_97449

Thank you Rob. I will contact the cluster adminstrator to resolve the problem then.

[Alan]

Sounds like permissions on the cluster: ie the host can't see/connect to the nodes.https://forum.ansys.com/discussion/comment/97449#Comment_97449

Dear Rob,n I just contacted the admin team. They said that access to the compute nodes is handled by the LSF scheduler and direct ssh access to the compute nodes is not allowed. If my simulation requires ssh type of access, the 'blaunch' (https://www.ibm.com/support/knowledgecenter/SSWRJV_10.1.0/lsf_command_ref/blaunch.8.html) command should provide the same type of functionality. nI was trying to use blaunch for fluent but can not find any guides from ANSYS Learning Forum. Do you by chance have any information about integrating blaunch with Fluent?.thanks!nAlann

[Rob]

One for then: I just break clusters.... n

[ANSYS_MMadore]

For blaunch to work passwordless you need to set these variables:nPlease set the below two system environment variablesnIn Bash Shell:nexport FLUENT_SSH=blaunchnexport SCHEDULER_RSH=1nnIn C Shell:nsetenv FLUENT_SSH blaunchnsetenv SCHEDULER_RSH 1nnAlso, you have to use -scheduler_tight_coupling in your command line.nnThanksnMattn

[Alan]

@Alan For blaunch to work passwordless you need to set these variables:Please set the below two system environment variablesIn Bash Shell:export FLUENT_SSH=blaunchexport SCHEDULER_RSH=1In C Shell:setenv FLUENT_SSH blaunchsetenv SCHEDULER_RSH 1Also, you have to use -scheduler_tight_coupling in your command line.ThanksMatthttps://forum.ansys.com/discussion/comment/97620#Comment_97620

Thanks, Matt. nI tried your commands (In Bash Shell). I used the following command to launch Fluent:nfluent 3d -g -t4 -i ###.jou -scheduler_tight_couplingnIt does use the blaunch as the system shows: 'using remote shell blaunch'. But, it still comes up with an error message:nHost key verification failed.nError: It seems ssh is trying to verify authenticity of c39b12. Please resolve it and try again!nClient interrupted.nAny ideas to resolve this problem?.thanks in advance!nAla

[ANSYS_MMadore]

Could you please try adding n-mpi=openmpinnThanksnMattn

[Alan]

Could you please try adding -mpi=openmpiThanksMatthttps://forum.ansys.com/discussion/comment/97667#Comment_97667

-mpi=openmpi does not work. It says:nnA sequential map was requested, but not enough node entriesnwere given to support the requested number of processes:nn Num procs: 4n Num nodes: 1nnWe cannot continue - please adjust either the number of processesnor provide more node locations in the file.nnnBut, -mpi=intel works. But seems slower than default mpi (-mpi=ibmmpi) in the mpitest:nfor -mpi=intel with -scheduler_tight_couplingnPing pong latency test ...nping..pong..latency(usec)...count..hostnn0.....1.....0.300575........10000..0:c40b06n1.....2.....0.302362........10000..1:c40b06n2.....3.....0.307225........10000..2:c40b06n3.....0.....0.305513........10000..3:c40b06nnPing pong bandwidth test ...nping..pong..bandwidth(MB)...count.msg-size(MB)..hostnn0.....1.....8916.7..........10....4.............0:c40b06n1.....2.....8597.53.........10....4.............1:c40b06n2.....3.....8345.63.........10....4.............2:c40b06n3.....0.....8744.05.........10....4.............3:c40b06nnGlobal reduction test ...nMPI-function...time-per-msg(usec)..count...total-time(sec)..nnBcast..........0.89159.............5000....0.00445795.......nReduce.........1.11198.............5000....0.00555992.......nBarier.........0.34256.............5000....0.0017128........nnnFor -mpi=ibmmpi (which can not run Fluent because of the rsh problem)nPing pong latency test ...nping..pong..latency(usec)...count..hostnn0.....1.....0.2452..........10000..0:c40b06n1.....2.....0.2407..........10000..1:c40b06n2.....3.....0.23835.........10000..2:c40b06n3.....0.....0.248725........10000..3:c40b06nnPing pong bandwidth test ...nping..pong..bandwidth(MB)...count.msg-size(MB)..hostnn0.....1.....7351.01.........10....4.............0:c40b06n1.....2.....7340.88.........10....4.............1:c40b06n2.....3.....7287.95.........10....4.............2:c40b06n3.....0.....7446.61.........10....4.............3:c40b06nnGlobal reduction test ...nMPI-function...time-per-msg(usec)..count...total-time(sec)..nnBcast..........0.49243.............5000....0.00246215.......nReduce.........0.57559.............5000....0.00287795.......nBarier.........0.510406............5000....0.00255203.......nnIs the -mpi=intel a good option for running Fluent? or other -mpi options may be better?.

[ANSYS_MMadore]

Please try using -mpi=intelnnThanksnMattn