Problem with OpenMPI version on Ubuntu Linux

[Guilherme Fraga]

Hello,

I have been trying to run FDS in parallel on Ubuntu Linux. Since I've had some problems, I decided to start from zero, so I followed these steps:

1. I installed the most recent FDS version from the site.
2. Following the instructions (also from the site) for installing OpenMPI, I then used the command "sudo apt-get install openmpi-bin"
3. Following the instructions on the README file in the FDS installation directory, I then extracted the openmpi_1.8.4_linux_64.tar.gz file and altered the .bashrc_fds line (doing this without doing step 2 also did not achieve results).

However, when I try to run FDS in parallel, using the command provided in the User Guide, I receive messages that seem to indicate that the OpenMPI version is wrong (the errors read "received different version from [[63320,1],0]: 1.8.4 instead of 1.10.2"). I would like to know what is the problem and how to solve it.

Thanks

[Salah Benkorichi]

Try these following stpes : 

1. Install FDS , during the installation select

OpenMPI options
Press 1 to install OpenMPI manually [default]

Press 1 to select this option.

After installation,

Since you have not compiled openmpi yourself, at NIST, they provide openmpi with the bundle.

Try this : 

2. cd to FDS/FDS6/bin and then in terminal type:

 gunzip openmpi_1.8.4_linux_64.tar.gz
 tar -xvf openmpi_1.8.4_linux_64.tar

Open your .bashrc
copy these two lines. ( Don't forget to change usrname to your actual username, for example me: salah)

export MPIDIST_FDS=/home/usrname/FDS/FDS6/bin/openmpi_64

source ~/.bashrc_fds $MPIDIST_FDS

 

log off and log in

Now test to see the directory of mpirun

type :

which mpirun

you should get something like this

~/home/salah/FDS/FDS6/bin/openmpi_64/bin/mpirun

Try it and report what you get after it.

--
You received this message because you are subscribed to the Google Groups "FDS and Smokeview Discussions" group.
To unsubscribe from this group and stop receiving emails from it, send an email to fds-smv+unsubscribe@googlegroups.com.
To post to this group, send email to fds...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/fds-smv/6efcecd3-1329-42ef-84b1-3a51907822e0%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Guilherme Fraga]

Salah,


After following your instructions, I received the message that you said I should receive when typing "which mpirun". I proceeded then to try and run a parallel simulation, and now I'm receiving a different type of error:

"mpirun: error while loading shared libraries: libcrypto.so.10: cannot open shared object file: No such file or directory"

This is the same error I now receive when I use the command "mpirun --version".

To unsubscribe from this group and stop receiving emails from it, send an email to fds-smv+u...@googlegroups.com.

[Salah Benkorichi]

Ok, that means we're in the right direction. Try this now to fix that error

cd /lib/x86_64-linux-gnu
sudo ln -s libssl.so.1.0.0 libssl.so.10
sudo ln -s libcrypto.so.1.0.0 libcrypto.so.10

Test it now by : which mpirun 

and see if it gives back the right path, 

To unsubscribe from this group and stop receiving emails from it, send an email to fds-smv+unsubscribe@googlegroups.com.

To post to this group, send email to fds...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/fds-smv/74270c75-99d6-4bfe-bf65-ea8d6a45fbe3%40googlegroups.com.

[Guilherme Fraga]

Salah,

"which mpirun" provides the same path stated before. However, when I try to run a parallel fds simulation, I now receive a different error message (see the attached image). I should mention that the command I'm using is "mpirun -np 2 fds -hostfile my_hosts.txt test1.fds" and that the file my_hosts.txt exists in the same directory as test1.fds.

[Salah Benkorichi]

No, just run it as mpirun -np 2 fds test1.fds

Make sure that test1.fds has two meshe lines there.

On 17 October 2017 at 12:45, Guilherme Fraga <guilher...@gmail.com> wrote:

Salah,

"which mpirun" provides the same path stated before. However, when I try to run a parallel fds simulation, I now receive a different error message (see the attached image). I should mention that the command I'm using is "mpirun -np 2 fds -hostfile my_hosts.txt test1.fds" and that the file my_hosts.txt exists in the same directory as test1.fds.

--

You received this message because you are subscribed to the Google Groups "FDS and Smokeview Discussions" group.
To unsubscribe from this group and stop receiving emails from it, send an email to fds-smv+unsubscribe@googlegroups.com.
To post to this group, send email to fds...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/fds-smv/2d826bf4-7c9e-4eb1-9cac-30775dc7b009%40googlegroups.com.

[Guilherme Fraga]

Salah,

Thank you very much. Now the simulation is running without problems. However, I would like to know why I should use the command with the "-hostfile my_hosts.txt" part, since the manual specifies that the command to run a parallel simulation should include it. Additionally, in the mesh lines, should I keep the specification of a different MPI_PROCESS for each mesh?

[Salah Benkorichi]

I think "-hostfile my hosts.txt" is designed for people who use cluster.

2. For MPI_PROCESS is NO, You might one to do that in this case scenario for example: Say you have a computer with 3 processors, and you can run only 4 procesus on it (i.e. 4 meshes lines), and in your input file you have 10 meshes, then to avoid 6 meshes waiting until the other 4 are run, you can run all of them together as : 

&MESH IJK=10,10,10, XB=0.0112,0.01296,0.0,0.005,0.0,0.350, MPI_PROCESS = 0 /

&MESH IJK=10,10,10, XB=0.02636,0.02812,0.0,0.005,0.0,0.350, MPI_PROCESS = 0 /

&MESH IJK=10,10,10, XB=0.0112,0.01296,0.0,0.005,0.0,0.350, MPI_PROCESS = 1 /

&MESH IJK=10,10,10, XB=0.02636,0.02812,0.0,0.005,0.0,0.350, MPI_PROCESS = 1 /

&MESH IJK=10,10,10, XB=0.0112,0.01296,0.0,0.005,0.0,0.350, MPI_PROCESS = 2 /

&MESH IJK=10,10,10, XB=0.02636,0.02812,0.0,0.005,0.0,0.350, MPI_PROCESS = 2 /

&MESH IJK=10,10,10, XB=0.0112,0.01296,0.0,0.005,0.0,0.350, MPI_PROCESS = 2 /

&MESH IJK=10,10,10, XB=0.02636,0.02812,0.0,0.005,0.0,0.350, MPI_PROCESS = 2 /

&MESH IJK=10,10,10, XB=0.0112,0.01296,0.0,0.005,0.0,0.350, MPI_PROCESS = 3 /

&MESH IJK=10,10,10, XB=0.02636,0.02812,0.0,0.005,0.0,0.350, MPI_PROCESS = 3 /

On 17 October 2017 at 12:53, Guilherme Fraga <guilher...@gmail.com> wrote:

Salah,

Thank you very much. Now the simulation is running without problems. However, I would like to know why I should use the command with the "-hostfile my_hosts.txt" part, since the manual specifies that the command to run a parallel simulation should include it. Additionally, in the mesh lines, should I keep the specification of a different MPI_PROCESS for each mesh?

--

You received this message because you are subscribed to the Google Groups "FDS and Smokeview Discussions" group.
To unsubscribe from this group and stop receiving emails from it, send an email to fds-smv+unsubscribe@googlegroups.com.
To post to this group, send email to fds...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/fds-smv/0ba87c94-7c3f-4a17-9910-cc253b4f0176%40googlegroups.com.

[Guilherme Fraga]

Salah,

I`ve identified another problem, now when trying to compile FDS. Since the problem seems to me to be related to OpenMPI, I have decided to continue this thread.

I have downloaded the most recent FDS repository from github and tried to compile the code using gnu (I`m in a ubuntu linux machine). However, when I do this, I receive the following error message:

mpifort -c -O3 -cpp -DGITHASH_PP=\"-\" -DGITDATE_PP=\""\"" -DBUILDDATE_PP=\""Out 17, 2017  13:37:05\"" -DCOMPVER_PP=\""Gnu gfortran 5.4.0-6ubuntu1~16.04.4)"\" ../../Source/prec.f90
Cannot open configuration file /shared/openmpi_64/share/openmpi/mpifort-wrapper-data.txt
Error parsing data file mpifort: Not found
../makefile:112: recipe for target 'prec.o' failed
make: *** [prec.o] Error 243

Would you know a solution for this?

Thanks again for your help.

[Salah Benkorichi]

You want to use gnu, but you are trying to use mpifort there. Which is intel. That's now how you compile it. I've written on my blog how I did compiled on my machine. Follow these same steps and report what you get. Please follow them one by one, if an issue arises in between you should not proceed and report back.

FDS -Compilation using Gnu compiler on Linux (Ubuntu 16.04, Redhat 7/Centos 7)

1.Install the gfortran compiler.
In case you have other than 4.8 or 4.7 , you should uninstall it:
To know which version you have installed type :
$gfortran --version
in case you find it different, then type:
sudo apt-get remove gfortran
To install gfortran type :
sudo apt-get install gfortran-4.8

create symbolic link to gfortran :
sudo ln -sfn /usr/bin/gfortran-4.8 /usr/bin/gfortran
This would make gfortran in the same version being installed i.e. 4.8

2. Download and extract the latest version of Open MPI.
3. cd to the directory that you extracted Open MPI to.

4.Install Open MPI using the gnu compiler; configure and install using:

./configure CC=gcc --enable-mpi-fortran --enable-static --disable-shared -disable-dlopen --prefix=$Home/openmpi

(Note that the directory of my openmpi is in $/Home/openmpi)

then type

make
sudo make install

5. export some paths to your .bashrc file in your home directory:

export MPIDIST_ETH=$Home/openmpi
export LD_LIBRARY_PATH=$MPIDIST_ETH/lib
source ~/.bashrc_fds $MPIDIST_FDS

Note: MPIDIST_ETH= This should be pointing to the path you've chosen above. Mine was %Home/openmpi

save it and log off and log in again.

6. Build the FDS Linux version from the firemodels/fds repository:

cd Build/mpi_gnu_linux_64
./make_fds.sh

Test openmpi: which mpirun

Then test the compiled fds with mpirun. Example: mpirun -np 2 /home/salah/fds/Build/mpi_gnu_linux_64/fds_mpi_intel_linux_64 test.fds  --or you can simply copy the compiled fds to the fds input directory, and run it from there as : mpirun -np 2 ./fds-mpi_gnu_linux_64 test.fds 

Try it and report back.

--
You received this message because you are subscribed to the Google Groups "FDS and Smokeview Discussions" group.
To unsubscribe from this group and stop receiving emails from it, send an email to fds-smv+unsubscribe@googlegroups.com.
To post to this group, send email to fds...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/fds-smv/c09bae74-49c0-447c-ae07-134722bf2882%40googlegroups.com.

[Guilherme Fraga]

Salah

Should I install OpenMPI again? I have already installed it earlier using the command "sudo apt-get install openmpi-bin".

[Salah Benkorichi]

Yes, please follow the steps if you want to succeed in getting it working.

On 17 Oct 2017 5:23 pm, "Guilherme Fraga" <guilher...@gmail.com> wrote:

Salah

Should I install OpenMPI again? I have already installed it earlier using the command "sudo apt-get install openmpi-bin".

--

You received this message because you are subscribed to the Google Groups "FDS and Smokeview Discussions" group.
To unsubscribe from this group and stop receiving emails from it, send an email to fds-smv+unsubscribe@googlegroups.com.
To post to this group, send email to fds...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/fds-smv/27f04555-0ad6-40ab-8df0-138f8f1d2412%40googlegroups.com.

[Guilherme Fraga]

Salah,

I've followed all the steps you mentioned, without receiving any error messages in the process. However, when I try to build FDS, I still receive the same error message:

mpifort -c -O3 -cpp -DGITHASH_PP=\"-\" -DGITDATE_PP=\""\"" -DBUILDDATE_PP=\""Out 17, 2017  15:05:26\"" -DCOMPVER_PP=\""Gnu gfortran 4.8.5-4ubuntu2)"\" ../../Source/prec.f90

Cannot open configuration file /shared/openmpi_64/share/openmpi/mpifort-wrapper-data.txt
Error parsing data file mpifort: Not found
../makefile:112: recipe for target 'prec.o' failed
make: *** [prec.o] Error 243

Em terça-feira, 17 de outubro de 2017 14:25:13 UTC-2, Salah Benkorichi escreveu:

Yes, please follow the steps if you want to succeed in getting it working.

On 17 Oct 2017 5:23 pm, "Guilherme Fraga" <guilher...@gmail.com> wrote:

Salah

Should I install OpenMPI again? I have already installed it earlier using the command "sudo apt-get install openmpi-bin".

--
You received this message because you are subscribed to the Google Groups "FDS and Smokeview Discussions" group.

To unsubscribe from this group and stop receiving emails from it, send an email to fds-smv+u...@googlegroups.com.

[Salah Benkorichi]

Ok, looks like the compiler is not able to detect the mpifort, do me this: 

which ifort 

which mpifort 

What do they give?

On 17 October 2017 at 18:08, Guilherme Fraga <guilher...@gmail.com> wrote:

To unsubscribe from this group and stop receiving emails from it, send an email to fds-smv+unsubscribe@googlegroups.com.

To post to this group, send email to fds...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/fds-smv/974b22dd-35ab-426f-b3df-2ae6c95f191c%40googlegroups.com.

[Guilherme Fraga]

which ifort gives nothing, while which mpifort gives "/home/workstation/FDS/FDS6/bin/openmpi_64/bin/mpifort"

[Salah Benkorichi]

There you go. You are trying to target the mpifort that comes with the bundle instead of the compiled one. 

You need to comment the lines in your .bashrc file that you added to get the bundle working. 

Past here what you are putting in the .bashrc file regarding fds env

To unsubscribe from this group and stop receiving emails from it, send an email to fds-smv+unsubscribe@googlegroups.com.

To post to this group, send email to fds...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/fds-smv/c99a63d9-b440-49cf-9e56-caf63ece0b19%40googlegroups.com.

[Guilherme Fraga]

Salah,

Regarding the fds environment, I have the following lines in the .bashrc file:

#FDS -----------------------------------
export MPIDIST_FDS=/home/workstation/FDS/FDS6/bin/openmpi_64
export MPIDIST_ETH=/home/workstation/openmpi-3.0.0
export LD_LIBRARY_PATH=$MPIDIST_ETH/lib
export MPIDIST_IB=
INTEL_SHARED_LIB=$IFORT_COMPILER_LIB/intel64
source ~/.bashrc_fds $MPIDIST_FDS
#FDS -----------------------------------

[Salah Benkorichi]

Comment those lines, and use these lines. See if it works, if not, then I will retest it on my machine.

#FDS -----------------------------------
export MPIDIST_ETH=/home/workstation/openmpi-3.0.0
export LD_LIBRARY_PATH=$MPIDIST_ETH/lib
source ~/.bashrc_fds $MPIDIST_ETH
#FDS -----------------------------------

To unsubscribe from this group and stop receiving emails from it, send an email to fds-smv+unsubscribe@googlegroups.com.

To post to this group, send email to fds...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/fds-smv/cdc00d89-075f-4433-a2e2-8d83789ef72e%40googlegroups.com.

[Guilherme Fraga]

Salah,

It did not work. Additionally, I've tested again which ifort and wich mpifort, and now both give nothing.

[Salah Benkorichi]

Did you log off and log in again? Otherwise changes wont take effect.

To unsubscribe from this group and stop receiving emails from it, send an email to fds-smv+unsubscribe@googlegroups.com.

To post to this group, send email to fds...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/fds-smv/42c48be1-fa33-402c-8fea-36414247ab4c%40googlegroups.com.

[Guilherme Fraga]

An additional information, in case it may help: I've just realized that there is no lib directory inside the openmpi-3.0.0 directory. Shouldn't there be one, since the LD_LIBRARY_PATH is pointing to it?

[Salah Benkorichi]

Let me install it myself and check.

To unsubscribe from this group and stop receiving emails from it, send an email to fds-smv+unsubscribe@googlegroups.com.

To post to this group, send email to fds...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/fds-smv/d3bc4b97-6911-4feb-9d11-530aded577d2%40googlegroups.com.

[Guilherme Fraga]

Salah,

I have logged out and logged in, but the error persisted.

[Salah Benkorichi]

Hold on, I found the error, 

I will write to you some steps to follow.

To unsubscribe from this group and stop receiving emails from it, send an email to fds-smv+unsubscribe@googlegroups.com.

To post to this group, send email to fds...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/fds-smv/fc804e8d-62b2-49ae-993c-630ea414d915%40googlegroups.com.

[Salah Benkorichi]

I've noticed you're using the latest openmpi, follow these steps:

1. Create directory in your home called openmpi-3.0.0

2.  Download openmpi, then extract it. Like this :

$  tar -xzvf openmpi-3.0.0.tar.gz 

$ cd openmpi-3.0.0

3. Configure it as  :

./configure CC=gcc --enable-mpi-fortran --enable-static --disable-shared -disable-dlopen --prefix=/home/salah/openmpi-3.0.0/

NOTE: Change salah to your usrname in the line above  :

$ make

$ sudo make install

4. open your .bashrc file (make sure to comment or remove all other FDS env) and paste these lines:

export PATH=$PATH:$HOME/openmpi-3.0.0/bin

export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$HOME/openmpi-3.0.0/lib

source ~/.bashrc_fds $MPIDIST_FDS

log off and log in.

Let's debug it: $ which mpifort 

It should give something like this: /home/salah/openmpi-3.0.0/bin/mpifort

If not, do not proceed and report back.

If yes continue

$ cd to compilation directory (/mpi_gnu_linux_64)

$ ./make_fds.sh

See if this helps now. 

[Guilherme Fraga]

Salah,

Thanks very much for your reply. Unfortunately, I will not be able to test your solution today, because I don't have access to my computer. Tomorrow I will test it and report back.

[Guilherme Fraga]

Salah,

I've tried your new procedure. Three things are worth noting:

1. which mpifort gives, in fact, "/home/workstation/openmpi-3.0.0/bin/mpifort", as expected.
2. Bulding the code does not give the same error as before. Building continues almost through to the end (from the prompt, I see that all modules were processed), however, before finishing, a new message appears:

/usr/bin/ld: /home/workstation/openmpi-3.0.0/lib/libopen-pal.a(pdl_pdlopen_module.o): undefined reference to symbol 'dlclose@@GLIBC_2.2.5'
//lib/x86_64-linux-gnu/libdl.so.2: error adding symbols: DSO missing from command line
collect2: error: ld returned 1 exit status
../makefile:319: recipe for target 'mpi_gnu_linux_64' failed
make: *** [mpi_gnu_linux_64] Error 1

3. I can no longer run a fds parallel simulation (using the default executable). I receive the same message mentioned in my first post in this thread:

[workstation-lrt:25937] [[46336,0],0] tcp_peer_recv_connect_ack: received different version from [[46336,1],0]: 1.8.4 instead of 1.10.2
[workstation-lrt:25937] [[46336,0],0] tcp_peer_recv_connect_ack: received different version from [[46336,1],1]: 1.8.4 instead of 1.10.2

[Salah Benkorichi]

Yes, I can see it. I was checking it now, there seems to be a bug. We need to carry it out from now on the issue tracker. Please post an issue here: https://github.com/firemodels/fds/issues

We'll take it from there. 

Salah

--

You received this message because you are subscribed to the Google Groups "FDS and Smokeview Discussions" group.
To unsubscribe from this group and stop receiving emails from it, send an email to fds-smv+unsubscribe@googlegroups.com.
To post to this group, send email to fds...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/fds-smv/8da9bb4f-c186-4bd3-bda8-4f546b9efaa7%40googlegroups.com.

[Salah Benkorichi]

I you don't know how to open an issue, let me know.

[Guilherme Fraga]

Salah,

I've submitted an issue (#5586), in which I described the problems and procedures followed by me. I've never opened an issue before, so I hope I did it correctly. Thanks very much for your help.

Em quarta-feira, 18 de outubro de 2017 08:38:30 UTC-2, Salah Benkorichi escreveu:

I you don't know how to open an issue, let me know.

On 18 October 2017 at 11:17, Salah Benkorichi <benkori...@gmail.com> wrote:

Yes, I can see it. I was checking it now, there seems to be a bug. We need to carry it out from now on the issue tracker. Please post an issue here: https://github.com/firemodels/fds/issues

We'll take it from there. 

Salah

On 18 October 2017 at 11:10, Guilherme Fraga <guilher...@gmail.com> wrote:

Salah,

I've tried your new procedure. Three things are worth noting:

1. which mpifort gives, in fact, "/home/workstation/openmpi-3.0.0/bin/mpifort", as expected.
2. Bulding the code does not give the same error as before. Building continues almost through to the end (from the prompt, I see that all modules were processed), however, before finishing, a new message appears:

/usr/bin/ld: /home/workstation/openmpi-3.0.0/lib/libopen-pal.a(pdl_pdlopen_module.o): undefined reference to symbol 'dlclose@@GLIBC_2.2.5'
//lib/x86_64-linux-gnu/libdl.so.2: error adding symbols: DSO missing from command line
collect2: error: ld returned 1 exit status
../makefile:319: recipe for target 'mpi_gnu_linux_64' failed
make: *** [mpi_gnu_linux_64] Error 1

3. I can no longer run a fds parallel simulation (using the default executable). I receive the same message mentioned in my first post in this thread:

[workstation-lrt:25937] [[46336,0],0] tcp_peer_recv_connect_ack: received different version from [[46336,1],0]: 1.8.4 instead of 1.10.2
[workstation-lrt:25937] [[46336,0],0] tcp_peer_recv_connect_ack: received different version from [[46336,1],1]: 1.8.4 instead of 1.10.2


Em terça-feira, 17 de outubro de 2017 19:02:58 UTC-2, Guilherme Fraga escreveu:

Salah,

Thanks very much for your reply. Unfortunately, I will not be able to test your solution today, because I don't have access to my computer. Tomorrow I will test it and report back.

--
You received this message because you are subscribed to the Google Groups "FDS and Smokeview Discussions" group.

To unsubscribe from this group and stop receiving emails from it, send an email to fds-smv+u...@googlegroups.com.

To post to this group, send email to fds...@googlegroups.com.

[Salah Benkorichi]

No problem. Just keep in track with the discussion until we fix it.

Thanks.

To unsubscribe from this group and stop receiving emails from it, send an email to fds-smv+unsubscribe@googlegroups.com.

To post to this group, send email to fds...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/fds-smv/a5b8a077-01de-42df-96a1-21a0fe61fbde%40googlegroups.com.

[Glenn Forney]

Running installed fds and fds built in repo is tricky. The startup files need to point to the openmpi dist that was used to build each. If your startup files no longer point to openmpi that came with installed fds then installed fds won't work. So if you switch from one to the other you need to change your startup files accordingly.

To view this discussion on the web visit https://groups.google.com/d/msgid/fds-smv/CABK0Y1mkPzX1ROfUPt2CT5ap50MNzuipUbOrVrF69EoqoJkwvA%40mail.gmail.com.

[Guilherme Fraga]

Glenn,

When you say "startup files" are you referring to the lines in the .bashrc file?

To view this discussion on the web visit https://groups.google.com/d/msgid/fds-smv/a5b8a077-01de-42df-96a1-21a0fe61fbde%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "FDS and Smokeview Discussions" group.

To unsubscribe from this group and stop receiving emails from it, send an email to fds-smv+u...@googlegroups.com.
To post to this group, send email to fds...@googlegroups.com.

[Salah Benkorichi]

Yes

To unsubscribe from this group and stop receiving emails from it, send an email to fds-smv+unsubscribe@googlegroups.com.

To post to this group, send email to fds...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/fds-smv/e54dabcb-3583-40ca-aa18-ce1871e3715a%40googlegroups.com.

[Guilherme Fraga]

Ok. I see no problem in altering the .bashrc file from building FDS to executing FDS. In fact, the executing part I am able to do without problems, following what was initially suggested by Salah. However, even after using OpenMPI 2.1.0 (as suggested in issue #5586) and modifying the .bashrc file accordingly, I still can't build the code.

Yes

[Salah Benkorichi]

Please report on the issue tracker, and show the error you received after you compiled 2.1.0 

which mpifort : What does it give?

which mpirun: What does it give?

To unsubscribe from this group and stop receiving emails from it, send an email to fds-smv+unsubscribe@googlegroups.com.

To post to this group, send email to fds...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/fds-smv/5a529a67-4411-4ace-afda-d739c299e1a3%40googlegroups.com.

[Glenn Forney]

yes , any lines in .bashrc, .bash_profile etc that add entries to  PATH or LD_LIBRARY_PATH variables  for the OpenMPI distdribution you are using

To unsubscribe from this group and stop receiving emails from it, send an email to fds-smv+unsubscribe@googlegroups.com.

To post to this group, send email to fds...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/fds-smv/e54dabcb-3583-40ca-aa18-ce1871e3715a%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--

Glenn Forney