Now change the compiler to something else from Cray and run again.

On Dec 24, 2014, at 12:56 AM, Satish Balay <[email protected]> wrote:

wrt cray modules - I see the following env variables set [with a 'MODULE' in the env variable name]

Satish

------- balay@hopper01:~> env |grep MODULE MODULE_VERSION_STACK=3.2.6.7 LIBRARYMODULES=acml:alps:apprentice2:atp:cray-fftw:cray-libsci:cray-mpich2:cray-petsc:cray-petsc-complex:cray-shmem:cray-tpsl:cray-trilinos:cudatoolkit:fftw:ga:hdf5:hdf5-parallel:iobuf:lgdb:libfast:libsci_acc:mpich1:mpich2:mrnet:netcdf:netcdf-hdf5parallel:netcdf-nofsync:netcdf-nofsync-hdf5parallel:ntk:onesided:papi:parallel-netcdf:petsc:petsc-complex:pmi:shmem:tpsl:trilinos:xt-atp:xt-lgdb:xt-libsci:xt-mpt:xt-papi:/etc/opt/cray/modules/site_librarymodules USERMODULES=acml:alps:apprentice2:atp:blcr:cce:chapel:cray-fftw:cray-libsci:cray-mpich2:craypat:craype:cray-petsc:cray-petsc-complex:cray-shmem:cray-tpsl:cray-trilinos:cudatoolkit:ddt:fftw:ga:gcc:hdf5:hdf5-parallel:intel:iobuf:java:lgdb:libfast:libsci_acc:mpich1:mrnet:netcdf:netcdf-hdf5parallel:netcdf-nofsync:netcdf-nofsync-hdf5parallel:ntk:onesided:papi:parallel-netcdf:pathscale:perftools:petsc:petsc-complex:pgi:pmi:PrgEnv-cray:PrgEnv-gnu:PrgEnv-intel:PrgEnv-pathscale:PrgEnv-pgi:stat:totalview:tpsl:trilinos:xt-asyncpe:xt-craypat:xt-lgdb:xt-libsci:xt-mpich2:xt-mpt:xt-papi:xt-shmem:xt-totalview:/etc/opt/cray/modules/site_usermodules:/usr/common/nsg/Modules/modulefiles MODULE_VERSION=3.2.6.7 TARGETMODULES=craype-abudhabi:craype-abudhabi-cu:craype-accel-nvidia20:craype-accel-nvidia30:craype-accel-nvidia35:craype-barcelona:craype-hugepages128K:craype-hugepages128M:craype-hugepages16M:craype-hugepages256M:craype-hugepages2M:craype-hugepages512K:craype-hugepages512M:craype-hugepages64M:craype-hugepages8M:craype-interlagos:craype-interlagos-cu:craype-istanbul:craype-ivybridge:craype-knc:craype-mc12:craype-mc8:craype-network-aries:craype-network-gemini:craype-network-seastar:craype-sandybridge:craype-shanghai:craype-target-compute_node:craype-target-local_host:craype-target-native:craype-target-petest:craype-xeon:xtpe-barcelona:xtpe-interlagos:xtpe-interlagos-cu:xtpe-istanbul:xtpe-mc12:xtpe-mc8:xtpe-network-gemini:xtpe-network-seastar:xtpe-shanghai:xtpe-target-native:xtpe-xeon:/etc/opt/cray/modules/site_targetmodules MODULEPATH=/opt/cray/gem/modulefiles:/opt/cray/xt-asyncpe/default/modulefiles:/opt/cray/modulefiles:/opt/modulefiles:/usr/common/usg/Modules/modulefiles:/usr/syscom/nsg/modulefiles:/usr/syscom/nsg/opt/modulefiles:/usr/common/acts/Modules/modulefiles:/usr/common/ftg/Modules/modulefiles:/usr/common/graphics/Modules/modulefiles:/usr/common/jgi/Modules/modulefiles:/usr/common/tig/Modules/modulefiles LOADEDMODULES=modules/3.2.6.6:nsg/1.2.0:modules/3.2.6.7:xtpe-network-gemini:xt-asyncpe/5.23:pgi/13.6.0:cray-libsci/12.1.01:udreg/2.3.2-1.0402.7546.1.5.gem:ugni/5.0-1.0402.7551.1.10.gem:pmi/4.0.1-1.0000.9753.86.3.gem:dmapp/4.0.1-1.0402.7784.4.1.gem:gni-headers/2.1-1.0402.7541.1.5.gem:xpmem/0.1-2.0402.45248.1.5.gem:job/1.5.5-0.1_2.0402.45272.1.5.gem:csa/3.0.0-1_2.0402.45268.1.90.gem:rca/1.0.0-2.0402.47290.7.1.gem:audit/1.0.0-1.0402.45273.1.86.gem:ccm/2.2.0-1.0402.46086.4.120:configuration/1.0-1.0402.45284.1.2.gem:hosts/1.0-1.0402.45251.1.86.gem:lbcd/2.1-1.0402.45245.1.2.gem:nodehealth/5.1-1.0402.45895.3.76.gem:pdsh/2.26-1.0402.45278.1.1.gem:shared-root/1.0-1.0402.46893.3.17.gem:switch/1.0-1.0402.45840.2.63.gem:xe-sysroot/4.2.34:atp/1.7.0:PrgEnv-pgi/4.2.34:eswrap/1.0.20-1.010102.662.0:craype-mc12:cray-shmem/6.0.1:cray-mpich/6.0.1:torque/4.2.7.h1:moab/7.2.7-e7c070d1-b3-SUSE11:darshan/2.3.0:usg-default-modules/1.1 PRGENVMODULES=PrgEnv-cray:PrgEnv-gnu:PrgEnv-intel:PrgEnv-pathscale:PrgEnv-pgi TOOLMODULES=apprentice:apprentice2:atp:chapel:craypat:ddt:gdb:iobuf:mrnet:papi:perftools:stat:totalview:xt-craypat:xt-lgdb:xt-papi:xt-totalview:/etc/opt/cray/modules/site_toolmodules MODULESHOME=/opt/modules/3.2.6.7

But parsing this and figuring out - which module change is significant [wrt petsc build] and which one is not - is perhaps not easy. [Perhaps this is what Jed is refering to..]

Satish

On Tue, 23 Dec 2014, Barry Smith wrote:

...

I couldn't find any environmental variables for softenv and haven't logged into a cray in decades. We don't have to abort for all changes but perhaps could produce warning text or detect serious changes?

Barry

...

On Dec 23, 2014, at 9:31 PM, Jed Brown <[email protected]> wrote:

Barry Smith <[email protected]> writes:

...

Redirecting Jed's question specifically for PETSc configure.

Can we/should we save the modules setting at configure time and then check them always at make time? Is some environmental variable set that has a unique value based on the modules loaded?

We can do it for modules (like Cray) and softenv, but aborting in all cases of environment change is not tractable and would be too much noise.

...

If this is a common problem for us then we should be doing this.

It's a common problem in the sense that a lot of users get confused over this and it generates a significant amount of support email.