Alas, unable to even make.

joseph...@gmail.com

unread,

Jun 23, 2019, 4:02:35 PM6/23/19

to moltres-users

I followed the instructions and got this output:

/bin/sh: 1: /home/fusternino/moltres/moose/libmesh/installed/contrib/bin/libmesh-config: not found

/bin/sh: 1: /home/fusternino/moltres/moose/libmesh/installed/libtool: not found

/bin/sh: 1: /home/fusternino/moltres/moose/libmesh/installed/contrib/bin/libmesh-config: not found

Compiling C++ (in opt mode) /home/fusternino/moltres/moose/framework/build/unity_src/src_Unity.C...

/bin/sh: 1: /home/fusternino/moltres/moose/libmesh/installed/libtool: not found

/bin/sh: 1: /home/fusternino/moltres/moose/libmesh/installed/contrib/bin/libmesh-config: not found

/home/fusternino/moltres/moose/framework/build.mk:97: recipe for target '/home/fusternino/moltres/moose/framework/build/unity_src/src_Unity..opt.lo' failed

make: *** [/home/fusternino/moltres/moose/framework/build/unity_src/src_Unity..opt.lo] Error 127

Any advice? I do notice that my actual MOOSE directory has all of these while the (supposedly) updated submodule does not.

From looking through some of the files, the error may be due to a previous installation of llapack? Not exactly obvious.

Mateusz Pater

unread,

Jun 24, 2019, 4:45:59 AM6/24/19

to moltre...@googlegroups.com

Hi,

Did you run the update_and_rebuild_libmesh script?

If I remember correctly the problem in my case was in some of the Makefile scripts which look for Moltres inside Moose repository, whereas the instruction said the Moltres repository should be a separate one. Try checking where Libmesh searches for the config files and if they are really there.

Kind regards,

Matt

joseph...@gmail.com

unread,

Jun 24, 2019, 4:52:28 AM6/24/19

to moltres-users

I had actually tried that also. I don't have the output (and I remember it taking too long to just quickly get it) but I remember it failed to make and then I did ./run_tests and all failed but one.

So put Moltres inside Moose? Sorry I'm new to this whole thing-best practice is generally then to have them all at the same directory level? I did that with Squirrel and Zapdos and no issues. Will try again soon.

Alexander Lindsay

unread,

Jun 24, 2019, 8:25:41 AM6/24/19

to moltre...@googlegroups.com

1) cd into the root moltres directory

2) git submodule update —init —recursive

3) cd moose (so you should be in moltres/moose now)

4) scripts/update_and_rebuild_libmesh.sh

5) cd .. (now should be back in moltres root)

6) make (append -j option with the number of processors available on your machine)

You can also build with moose and moltres on the same level but I would encourage use of the submodule.

--
You received this message because you are subscribed to the Google Groups "moltres-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to moltres-user...@googlegroups.com.
To post to this group, send email to moltre...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/moltres-users/8aaf44ce-070d-4144-9d65-07b53430bd33%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

joseph...@gmail.com

unread,

Jun 27, 2019, 4:24:34 AM6/27/19

to moltres-users

I tried this . . . current error is now:

configure: error: *** PETSc was not found, but --enable-petsc-required was specified.

make: *** No targets specified and no makefile found. Stop.

I then tried running ./update_and_rebuild_petsc.sh but that also failed:

makefile:15: arch-moose/lib/petsc/conf/petscvariables: No such file or directory

/home/fusternino/moltres/moose/scripts/../petsc/lib/petsc/conf/variables:119: /home/fusternino/moltres/moose/scripts/../petsc/arch-moose/lib/petsc/conf/petscvariables: No such file or directory

/home/fusternino/moltres/moose/scripts/../petsc/lib/petsc/conf/rules:960: /home/fusternino/moltres/moose/scripts/../petsc/arch-moose/lib/petsc/conf/petscrules: No such file or directory

make: *** No rule to make target '/home/fusternino/moltres/moose/scripts/../petsc/arch-moose/lib/petsc/conf/petscrules'. Stop.

On Monday, June 24, 2019 at 5:25:41 AM UTC-7, Alexander Lindsay wrote:

1) cd into the root moltres directory
2) git submodule update —init —recursive
3) cd moose (so you should be in moltres/moose now)
4) scripts/update_and_rebuild_libmesh.sh
5) cd .. (now should be back in moltres root)
6) make (append -j option with the number of processors available on your machine)

You can also build with moose and moltres on the same level but I would encourage use of the submodule.

On Jun 24, 2019, at 2:52 AM, joseph...@gmail.com wrote:

I had actually tried that also. I don't have the output (and I remember it taking too long to just quickly get it) but I remember it failed to make and then I did ./run_tests and all failed but one.

So put Moltres inside Moose? Sorry I'm new to this whole thing-best practice is generally then to have them all at the same directory level? I did that with Squirrel and Zapdos and no issues. Will try again soon.

On Monday, June 24, 2019 at 1:45:59 AM UTC-7, Mateusz Pater wrote:
Hi,

Did you run the update_and_rebuild_libmesh script?

If I remember correctly the problem in my case was in some of the Makefile scripts which look for Moltres inside Moose repository, whereas the instruction said the Moltres repository should be a separate one. Try checking where Libmesh searches for the config files and if they are really there.

Kind regards,
Matt

--
You received this message because you are subscribed to the Google Groups "moltres-users" group.

To unsubscribe from this group and stop receiving emails from it, send an email to moltre...@googlegroups.com.

Alexander Lindsay

unread,

Jun 27, 2019, 2:47:12 PM6/27/19

to moltre...@googlegroups.com, moose...@googlegroups.com

I’m cc’ing the moose users list...

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

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

To view this discussion on the web visit https://groups.google.com/d/msgid/moltres-users/041321cb-ffba-4a88-9b4c-c3fda29719a9%40googlegroups.com.

Alexander Lindsay

unread,

Jun 28, 2019, 12:04:03 AM6/28/19

to moltres-users

I think we're going to need to see the entire output of the commands that you're running. There is simply not enough information here. If I had to guess, I would guess that you're not actually able to clone the petsc repository. If you cd into the petsc directory, and type `ls` what do you see?

Try running this:

scripts/update_and_rebuild_petsc.sh 2>&1 | tee build.log

and then attach your build.log file here to the list.

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

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

To view this discussion on the web visit https://groups.google.com/d/msgid/moltres-users/041321cb-ffba-4a88-9b4c-c3fda29719a9%40googlegroups.com.

Joseph Fustero

unread,

Jun 29, 2019, 5:06:25 AM6/29/19

to moltre...@googlegroups.com

In the petsc folder I see:

arch-moose config CONTRIBUTING include lib RDict.log share
bitbucket-pipelines.yml configure gmakefile interfaces LICENSE README.md src
CODE_OF_CONDUCT.md configure.log gmakefile.test Jenkinsfile makefile setup.py systems

I attached the build.log and also configure.log which is referenced near the end of build.log.

You received this message because you are subscribed to a topic in the Google Groups "moltres-users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/moltres-users/tx7ginP3nKY/unsubscribe.
To unsubscribe from this group and all its topics, send an email to moltres-user...@googlegroups.com.

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

To view this discussion on the web visit https://groups.google.com/d/msgid/moltres-users/CANFcJrEY_TAUnhKYjH11uS4rMB_vRGDwnT4VZhOLpp4pyD_DJQ%40mail.gmail.com.

build.log

configure.log

Alexander Lindsay

unread,

Jun 29, 2019, 9:27:37 PM6/29/19

to moltres-users

So there are linker errors like this:

Possible ERROR while running linker: exit code 1
stderr:
/home/fusternino/moltres/moose/petsc/arch-moose/lib/libsuperlu_dist.so: undefined reference to `ompi_mpi_op_sum'
/home/fusternino/moltres/moose/petsc/arch-moose/lib/libsuperlu_dist.so: undefined reference to `ompi_mpi_op_min'
/home/fusternino/moltres/moose/petsc/arch-moose/lib/libsuperlu_dist.so: undefined reference to `ompi_mpi_comm_world'
/home/fusternino/moltres/moose/petsc/arch-moose/lib/libsuperlu_dist.so: undefined reference to `ompi_mpi_op_max'
/home/fusternino/moltres/moose/petsc/arch-moose/lib/libsuperlu_dist.so: undefined reference to `ompi_mpi_double'
/home/fusternino/moltres/moose/petsc/arch-moose/lib/libsuperlu_dist.so: undefined reference to `ompi_mpi_float'
/home/fusternino/moltres/moose/petsc/arch-moose/lib/libparmetis.so: undefined reference to `ompi_mpi_2int'
/home/fusternino/moltres/moose/petsc/arch-moose/lib/libparmetis.so: undefined reference to `ompi_mpi_op_minloc'
/home/fusternino/moltres/moose/petsc/arch-moose/lib/libsuperlu_dist.so: undefined reference to `ompi_mpi_comm_null'
/home/fusternino/moltres/moose/petsc/arch-moose/lib/libsuperlu_dist.so: undefined reference to `ompi_mpi_int'
/home/fusternino/moltres/moose/petsc/arch-moose/lib/libparmetis.so: undefined reference to `MPI_Comm_f2c'
/home/fusternino/moltres/moose/petsc/arch-moose/lib/libsuperlu_dist.so: undefined reference to `ompi_request_null'
/home/fusternino/moltres/moose/petsc/arch-moose/lib/libparmetis.so: undefined reference to `ompi_mpi_double_int'
/home/fusternino/moltres/moose/petsc/arch-moose/lib/libsuperlu_dist.so: undefined reference to `ompi_mpi_datatype_null'
collect2: error: ld returned 1 exit status

It looks like you must have a system MPI which would make sense since I see things in your log about OpenFOAM and other things. Are you loading the moose environment? If so, then I can understand the linker problems above. I would advise unloading the MOOSE environment since it appears like you already have your own MPI stack, and then try rebuilding petsc using our script.

You can look in your .bashrc or .bash_profile to ensure that you are not loading the moose environment. Then I would start a fresh terminal, and run `which mpicc` and make sure that it's not coming from /opt/moose.

To view this discussion on the web visit https://groups.google.com/d/msgid/moltres-users/CAEuUXEVa2i3fnfrJCV9uXATgE_oM%3D-eCgqTAjt3kmQF-h8p9OQ%40mail.gmail.com.

Joseph Fustero

unread,

Jun 30, 2019, 5:17:22 AM6/30/19

to moltre...@googlegroups.com

So, I unloaded both MOOSE as well as OpenFOAM. I then update_and_rebuild_petsc which was successful and then I was prompted run:

make PETSC_DIR=/home/fusternino/moltres/moose/scripts/../petsc PETSC_ARCH=arch-moose check

However, this doesn't work and needs to be changed to:

make PETSC_DIR=/home/fusternino/moltres/moose/petsc PETSC_ARCH=arch-moose check

This went well. I then ran update_and_rebuild_libmesh which went well.

I then did make -j8 from the root directory and that was good although very laggy. No errors, though.

I then did make_tests but was prompted that my python needed to be 2.7 so I unloaded 3.7 (I use Anaconda3) and that went well.

Note that of the 13 tests, 12 were passed. Note that the first test to run, twod_axi_coupled.coupled, while it did not fail, it was however skipped.

Thanks for helping me out!

To view this discussion on the web visit https://groups.google.com/d/msgid/moltres-users/CANFcJrHGc4gkL5rS_F4Gue%2Bh15MVvsbPjoAH%3DmxAtCXDbpbzYQ%40mail.gmail.com.

Alexander Lindsay

unread,

Jun 30, 2019, 9:13:42 AM6/30/19

to moltre...@googlegroups.com

That’s great! I’m glad we got it to work!

To view this discussion on the web visit https://groups.google.com/d/msgid/moltres-users/CAEuUXEXWSWVtQ6P3yRWyUp%3D6DbomMS5Xbbt2i8AJ9%2BUrb8qK-w%40mail.gmail.com.

Reply all

Reply to author

Forward