MPI ATS Installation problem

445 views
Skip to first unread message

James Thornton

unread,
Aug 5, 2023, 9:06:37 AM8/5/23
to Amanzi-ATS Users
Hello, 

I'm trying to install ATS for the first time on MacOS 13.4 (M2), but I'm having a problem.

Based on previous posts I've tested several combinations for mpi_root_dir, including: 

/usr/local/bin/mpicc
/usr/local/bin
/usr/local/

I've confirmed that mpicc, mpifort and mpicxx are all in the "bin" directory. Example logs are attached. 

I've tried Anaconda and Brew distributions of open mpi, as well as adding --with-fort-flags="-fallow-argument-mismatch" --with-c-flags="-Wno-error-implicit-function-declaration" to the build_ATS_generic.sh call  

Elsewhere I see an error "The Open MPI wrapper compiler was unable to find the specified compiler arm64-apple-darwin20.0.0-clang++ in your PATH", which sounds like it could be relevant, but which unfortunately I am unable to fix. 

Many thanks indeed for any help!

Best wishes,

James
log1.txt
log2.txt

Amanzi-ATS Users

unread,
Aug 5, 2023, 3:43:59 PM8/5/23
to Amanzi-ATS Users
So brew installs the mpi wrappers in one directory and the mpi libraries in another.  Try running:

mpicc -show

This should include some part about:

-I /usr/local/homebrew/Cellar/open-mpi/###/include

You should set --with-mpi= /usr/local/homebrew/Cellar/open-mpi/### on your bootstrap command.

Ethan

JM Thornton

unread,
Aug 5, 2023, 4:42:56 PM8/5/23
to Amanzi-ATS Users
Thanks Ethan, 

That (/opt/homebrew/Cellar/open-mpi/4.1.5 in my case) seems to have done the trick.

Best wishes, 

James

--
You received this message because you are subscribed to a topic in the Google Groups "Amanzi-ATS Users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/ats-users/Nfl9HP3Hhns/unsubscribe.
To unsubscribe from this group and all its topics, send an email to ats-users+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ats-users/a79ee013-8870-4a9e-9e01-676ae5ba91cbn%40googlegroups.com.

James Thornton

unread,
Aug 6, 2023, 7:23:26 AM8/6/23
to Amanzi-ATS Users
Unfortunately I then encounter another error:

[ 10%] Performing build step for 'boost'
CMake Error at /Users/james/Downloads/amanzi-ats/amanzi_tpls-build-master-Release/boost/boost-timestamps/boost-build-Release.cmake:49 (message):
  Command failed: 1

   '/Applications/CMake.app/Contents/bin/cmake' '-P' '/Users/james/Downloads/amanzi-ats/amanzi_tpls-build-master-Release/boost/boost-build-step.cmake'

  See also

    /Users/james/Downloads/amanzi-ats/amanzi_tpls-build-master-Release/boost/boost-timestamps/boost-build-*.log


make[2]: *** [boost/boost-timestamps/boost-build] Error 1
make[1]: *** [CMakeFiles/boost.dir/all] Error 2
make: *** [all] Error 2
Amanzi Bootstrap ERROR: [31m Failed to build TPLs [m



boost-build-out.log is empty.

Any help would be greatly appreciated!

Best wishes,

James
log.txt
boost-build-err.log

Coon, Ethan

unread,
Aug 6, 2023, 8:46:49 AM8/6/23
to James Thornton, Amanzi-ATS Users
Yuck, we’ve been having boost problems lately, and intend to stop using it soon.  But for now, can you send the full output (forward the screen output to a file) and both the boost-build-out.log and boost-build-err.log, plus boost-config-{out,err}.log as well.  Sometimes boost config errors are getting missed and then only show up when the build step fails.

Ethan

Sent from Ethan Coon’s phone.

From: 'James Thornton' via Amanzi-ATS Users <ats-...@googlegroups.com>
Sent: Sunday, August 6, 2023 7:23:25 AM
To: Amanzi-ATS Users <ats-...@googlegroups.com>
Subject: [EXTERNAL] Re: MPI ATS Installation problem
 
You received this message because you are subscribed to the Google Groups "Amanzi-ATS Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ats-users+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ats-users/b8ccb0db-385f-407a-9aa2-187e562af08cn%40googlegroups.com.

James Thornton

unread,
Aug 6, 2023, 9:06:14 AM8/6/23
to Amanzi-ATS Users
Thanks Ethan,

I already attached those files to the above message (log.txt and boost-build-err.log). boost-build-out.log was empty, so not attached. But don't hesitate to let me know if I should send something else. 

Also, I'm in no rush..getting everything sorted next week would also be fine. 

Best,

James

Coon, Ethan

unread,
Aug 6, 2023, 9:13:21 AM8/6/23
to James Thornton, Amanzi-ATS Users
Yep, this is a config problem.  Can you send the boost-config-{out,err}.log files?  

Ethan

Sent from Ethan Coon’s phone.
From: 'James Thornton' via Amanzi-ATS Users <ats-...@googlegroups.com>
Sent: Sunday, August 6, 2023 9:06:14 AM
To: Amanzi-ATS Users <ats-...@googlegroups.com>
Subject: Re: [EXTERNAL] Re: MPI ATS Installation problem
 

James Thornton

unread,
Aug 6, 2023, 9:18:34 AM8/6/23
to Amanzi-ATS Users
Here are those files.

Best,

James

On Sunday, 6 August 2023 at 15:13:21 UTC+2 Coon, Ethan wrote:
Yep, this is a config problem.  Can you send the boost-config-{out,err}.log files?  

Ethan

Sent from Ethan Coon’s phone.
boost-build-err.log

James Thornton

unread,
Aug 6, 2023, 9:23:31 AM8/6/23
to Amanzi-ATS Users
this one is empty. 

Ethan Coon

unread,
Aug 6, 2023, 9:33:42 AM8/6/23
to James Thornton, Amanzi-ATS Users
Whups, I think those are the build ones again.  In the same directory as those, you should also have a boost-config-out.log and boost-config-err.log file.  Each library like boost first gets configured, then gets built.  I suspect that there is an error in your configure step that is causing the build step to fail.

Ethan



--

-------------------------------------------------------------------
Ethan Coon
917-969-6831
https://www.ornl.gov/staff-profile/ethan-t-coon
-------------------------------------------------------------------

James Thornton

unread,
Aug 6, 2023, 9:57:47 AM8/6/23
to Amanzi-ATS Users
Sorry! It must be these ones:
boost-configure-out.log
boost-configure-err.log

Moulton, J. David

unread,
Aug 7, 2023, 1:44:52 AM8/7/23
to Amanzi-ATS Users, James Thornton
Hi James,

In those error logs it looks like there is some confusion coming from the compilers that are installed in Anaconda. I do use Anaconda to provide python3, but I've never had much luck with the mpi and compiler setup there. I think you were on the right track with the brew compilers, and ensuring that the brew directories are ahead of any Anaconda directories in your PATH.

If the mpi that comes with brew doesn't seem to work (I'm not sure which variants you have installed but mpi is likely built with the system gcc (which is really clang), and may point to gfortran from you compiler installation under brew). It probably doesn't use the gcc/g++ that you may have installed under brew, but you could check. Any way if this doesn't seem to work (i.e., doesn't pass our tests in the initial part of the CMake bootstrap process), then the most reliable thing is to build openmpi (or mpich) yourself using the compilers you really want to use (e.g., a complete and consistent set of gcc compilers).

Once you've got the compiler/mpi setup consistent and working, the builds should work without any issues.

Also, please double check that you have the Xcode command line tools installed. Recently we had an example of a build that was getting pretty far along without the command line tools, even though brew and macports both technically require them to get started effectively. This lack of command line tools generated some very misleading errors for sure, some of those were related to mpi compiler failures.

Cheers, David

________________________________________
From: 'James Thornton' via Amanzi-ATS Users <ats-...@googlegroups.com>
Sent: Sunday, August 6, 2023 7:57 AM
To: Amanzi-ATS Users
Subject: Re: [EXTERNAL] Re: MPI ATS Installation problem

Sorry! It must be these ones:

On Sunday, 6 August 2023 at 15:33:42 UTC+2 etc...@gmail.com wrote:
Whups, I think those are the build ones again. In the same directory as those, you should also have a boost-config-out.log and boost-config-err.log file. Each library like boost first gets configured, then gets built. I suspect that there is an error in your configure step that is causing the build step to fail.

Ethan

'/Applications/CMake.app/Contents/bin/cmake'<https://urldefense.com/v3/__http://CMake.app/Contents/bin/cmake'__;!!Bt8fGhp8LhKGRg!AKG-XzaCqTPYaaX44d5R19xDJrMlO7sLAMNHEt966ur8eT_HZB7LOD1TySz9Dh-nuVRKaC2eDYd0BX6geg8AHho$> '-P' '/Users/james/Downloads/amanzi-ats/amanzi_tpls-build-master-Release/boost/boost-build-step.cmake'
I've tried Anaconda and Brew distributions of open mpi, as well as adding --with-fort-flags="-fallow-argument-mismatch" --with-c-flags="-Wno-error-implicit-function-declaration" to the build_ATS_generic.sh<https://urldefense.com/v3/__http://build_ATS_generic.sh__;!!Bt8fGhp8LhKGRg!AKG-XzaCqTPYaaX44d5R19xDJrMlO7sLAMNHEt966ur8eT_HZB7LOD1TySz9Dh-nuVRKaC2eDYd0BX6g7zuHWpY$> call

Elsewhere I see an error "The Open MPI wrapper compiler was unable to find the specified compiler arm64-apple-darwin20.0.0-clang++ in your PATH", which sounds like it could be relevant, but which unfortunately I am unable to fix.

Many thanks indeed for any help!

Best wishes,

James

--
You received this message because you are subscribed to a topic in the Google Groups "Amanzi-ATS Users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/ats-users/Nfl9HP3Hhns/unsubscribe<https://urldefense.com/v3/__https://urldefense.us/v2/url?u=https-3A__groups.google.com_d_topic_ats-2Dusers_Nfl9HP3Hhns_unsubscribe&d=DwMFaQ&c=v4IIwRuZAmwupIjowmMWUmLasxPEgYsgNI-O7C4ViYc&r=Xw-NR7rAIqqldnLXRstTmg&m=mv6xoPRl0zwhpODSt9fxw7FY7RrP2amqUSdERdjTAij9u5RluZKg9VMXfzttLS0w&s=ZkIr-EOfNpCWYJGc1iyXzTcqEz_PkJqNLkmpBVL76HE&e=__;!!Bt8fGhp8LhKGRg!AKG-XzaCqTPYaaX44d5R19xDJrMlO7sLAMNHEt966ur8eT_HZB7LOD1TySz9Dh-nuVRKaC2eDYd0BX6gppOWIQ0$>.
To unsubscribe from this group and all its topics, send an email to ats-users+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ats-users/a79ee013-8870-4a9e-9e01-676ae5ba91cbn%40googlegroups.com<https://urldefense.com/v3/__https://urldefense.us/v2/url?u=https-3A__groups.google.com_d_msgid_ats-2Dusers_a79ee013-2D8870-2D4a9e-2D9e01-2D676ae5ba91cbn-2540googlegroups.com-3Futm-5Fmedium-3Demail-26utm-5Fsource-3Dfooter&d=DwMFaQ&c=v4IIwRuZAmwupIjowmMWUmLasxPEgYsgNI-O7C4ViYc&r=Xw-NR7rAIqqldnLXRstTmg&m=mv6xoPRl0zwhpODSt9fxw7FY7RrP2amqUSdERdjTAij9u5RluZKg9VMXfzttLS0w&s=FOJV4YVF86S2P0Rjs0hjsxUolGgXT0afSUZf9WUqpEo&e=__;!!Bt8fGhp8LhKGRg!AKG-XzaCqTPYaaX44d5R19xDJrMlO7sLAMNHEt966ur8eT_HZB7LOD1TySz9Dh-nuVRKaC2eDYd0BX6gASrzaZQ$>.

--
You received this message because you are subscribed to the Google Groups "Amanzi-ATS Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ats-users+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ats-users/b8ccb0db-385f-407a-9aa2-187e562af08cn%40googlegroups.com<https://urldefense.com/v3/__https://urldefense.us/v2/url?u=https-3A__groups.google.com_d_msgid_ats-2Dusers_b8ccb0db-2D385f-2D407a-2D9aa2-2D187e562af08cn-2540googlegroups.com-3Futm-5Fmedium-3Demail-26utm-5Fsource-3Dfooter&d=DwMFaQ&c=v4IIwRuZAmwupIjowmMWUmLasxPEgYsgNI-O7C4ViYc&r=Xw-NR7rAIqqldnLXRstTmg&m=mv6xoPRl0zwhpODSt9fxw7FY7RrP2amqUSdERdjTAij9u5RluZKg9VMXfzttLS0w&s=CmgN83xOxAAt8-TLrhtiQ5NcVfkcxv9sPA60m7-rtH0&e=__;!!Bt8fGhp8LhKGRg!AKG-XzaCqTPYaaX44d5R19xDJrMlO7sLAMNHEt966ur8eT_HZB7LOD1TySz9Dh-nuVRKaC2eDYd0BX6ga4EMy8M$>.

--
You received this message because you are subscribed to the Google Groups "Amanzi-ATS Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ats-users+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ats-users/c9edca9c-a6a9-4897-93d6-96ff745e7837n%40googlegroups.com<https://urldefense.com/v3/__https://urldefense.us/v2/url?u=https-3A__groups.google.com_d_msgid_ats-2Dusers_c9edca9c-2Da6a9-2D4897-2D93d6-2D96ff745e7837n-2540googlegroups.com-3Futm-5Fmedium-3Demail-26utm-5Fsource-3Dfooter&d=DwMFaQ&c=v4IIwRuZAmwupIjowmMWUmLasxPEgYsgNI-O7C4ViYc&r=Xw-NR7rAIqqldnLXRstTmg&m=dqXFGje-FPMJ6v5Hpyuf5Nq7XXxhtNyAbsB_OMHs-gEvqmdpOID_0BAi4Td32YYk&s=1BuT9nLRUsjp_gIwNTqgMP9zjJtQQz0QTraZPJoKHuU&e=__;!!Bt8fGhp8LhKGRg!AKG-XzaCqTPYaaX44d5R19xDJrMlO7sLAMNHEt966ur8eT_HZB7LOD1TySz9Dh-nuVRKaC2eDYd0BX6gcozOKg0$>.

--
You received this message because you are subscribed to the Google Groups "Amanzi-ATS Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ats-users+...@googlegroups.com<https://urldefense.com/v3/__http://tniopfoorp__;!!Bt8fGhp8LhKGRg!AKG-XzaCqTPYaaX44d5R19xDJrMlO7sLAMNHEt966ur8eT_HZB7LOD1TySz9Dh-nuVRKaC2eDYd0BX6gXtBURIc$>.
To view this discussion on the web visit https://groups.google.com/d/msgid/ats-users/0ff9e3d7-f0ac-4dd1-a235-b0ad5fb1962cn%40googlegroups.com<https://urldefense.com/v3/__https://groups.google.com/d/msgid/ats-users/0ff9e3d7-f0ac-4dd1-a235-b0ad5fb1962cn*40googlegroups.com?utm_medium=email&utm_source=footer__;JQ!!Bt8fGhp8LhKGRg!AKG-XzaCqTPYaaX44d5R19xDJrMlO7sLAMNHEt966ur8eT_HZB7LOD1TySz9Dh-nuVRKaC2eDYd0BX6gqTQc_pU$>.


--

-------------------------------------------------------------------
Ethan Coon
917-969-6831<tel:(917)%20969-6831>
https://www.ornl.gov/staff-profile/ethan-t-coon<https://urldefense.com/v3/__https://www.ornl.gov/staff-profile/ethan-t-coon__;!!Bt8fGhp8LhKGRg!AKG-XzaCqTPYaaX44d5R19xDJrMlO7sLAMNHEt966ur8eT_HZB7LOD1TySz9Dh-nuVRKaC2eDYd0BX6g1of8uso$>
-------------------------------------------------------------------

--
You received this message because you are subscribed to the Google Groups "Amanzi-ATS Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ats-users+...@googlegroups.com<mailto:ats-users+...@googlegroups.com>.
To view this discussion on the web visit https://groups.google.com/d/msgid/ats-users/82a64dea-82b8-49ff-b47c-aa6fca123457n%40googlegroups.com<https://urldefense.com/v3/__https://groups.google.com/d/msgid/ats-users/82a64dea-82b8-49ff-b47c-aa6fca123457n*40googlegroups.com?utm_medium=email&utm_source=footer__;JQ!!Bt8fGhp8LhKGRg!AKG-XzaCqTPYaaX44d5R19xDJrMlO7sLAMNHEt966ur8eT_HZB7LOD1TySz9Dh-nuVRKaC2eDYd0BX6ggjp1n40$>.

James Thornton

unread,
Aug 7, 2023, 10:06:38 AM8/7/23
to Amanzi-ATS Users
Hi David,

Many thanks, that was helpful.  

Ensuring brew was prioritised over Anaconda enabled the build to proceed well to 99%. 

There now seems to be a further error stemming from a missing file, models_transport_reg.hh.

I cloned the latest release (git clone -b amanzi-1.4.1 http://github.com/amanzi/amanzi $AMANZI_SRC_DIR). Should I adjust something?

Best wishes,

James

###############################################

In file included from /Users/james/Downloads/amanzi-ats/repos/amanzi/src/physics/ats/src/executables/main.cc:34:

/Users/james/Downloads/amanzi-ats/repos/amanzi/src/physics/ats/src/executables/ats_registration_files.hh:23:10: fatal error: 'models_transport_reg.hh' file not found

#include "models_transport_reg.hh"

         ^~~~~~~~~~~~~~~~~~~~~~~~~

In file included from /Users/james/Downloads/amanzi-ats/repos/amanzi/src/physics/ats/src/executables/test/Main.cc:16:

/Users/james/Downloads/amanzi-ats/repos/amanzi/src/physics/ats/src/executables/ats_registration_files.hh:23:10: fatal error: 'models_transport_reg.hh' file not found

#include "models_transport_reg.hh"

         ^~~~~~~~~~~~~~~~~~~~~~~~~

In file included from /Users/james/Downloads/amanzi-ats/repos/amanzi/src/physics/ats/src/executables/test/Main.cc:16:

/Users/james/Downloads/amanzi-ats/repos/amanzi/src/physics/ats/src/executables/ats_registration_files.hh:23:10: fatal error: 'models_transport_reg.hh' file not found

#include "models_transport_reg.hh"

         ^~~~~~~~~~~~~~~~~~~~~~~~~

3 warnings and 1 error generated.

3 warnings and 1 error generated.

make[2]: *** [src/physics/ats/src/executables/CMakeFiles/executable_coupled_water.dir/test/Main.cc.o] Error 1

make[2]: *** Waiting for unfinished jobs....

make[2]: *** [src/physics/ats/src/executables/CMakeFiles/executable_mesh_factory.dir/test/Main.cc.o] Error 1

make[2]: *** Waiting for unfinished jobs....

3 warnings and 1 error generated.

make[2]: *** [src/physics/ats/src/executables/CMakeFiles/ats.dir/main.cc.o] Error 1

make[1]: *** [src/physics/ats/src/executables/CMakeFiles/ats.dir/all] Error 2

make[1]: *** Waiting for unfinished jobs....

3 warnings generated.

make[1]: *** [src/physics/ats/src/executables/CMakeFiles/executable_mesh_factory.dir/all] Error 2

3 warnings generated.

make[1]: *** [src/physics/ats/src/executables/CMakeFiles/executable_coupled_water.dir/all] Error 2

make: *** [all] Error 2

Amanzi Bootstrap ERROR: Failed to build Amanzi

James Thornton

unread,
Aug 7, 2023, 3:29:44 PM8/7/23
to Amanzi-ATS Users
The latest master works fine (rather than the release). Everything is installed correctly now! 

Thanks again David and Ethan,

James

Coon, Ethan

unread,
Aug 8, 2023, 2:04:50 PM8/8/23
to James Thornton, Amanzi-ATS Users

Yep, I’m sorry, your trouble with 1.4 is my fault.  The branch ats-1.4 is correct and updated, but the submodule git hash listed in the amanzi-1.4 branch is inconsistent.  I will fix this now!


Ethan

 

Reply all
Reply to author
Forward
0 new messages