Better error messages and avoid check certificates.

[BFB]

…iles.xml

Introduces a new stub IAC
Adds a stub for Integrated Assessment Component (IAC).

Test suite: scripts_regression_tests.py for e3sm and cesm
Test baseline:
Test namelist changes:
Test status: [bit for bit, roundoff, climate changing]

Fixes [CIME Github issue #]

User interface changes?:

Update gh-pages html (Y/N)?:

Code review: jedwards

Changes the config files in CIME to reflect recent changes in Livermore
Computing resources:

Update syrah and quartz machine config entries to explicitly load
mvapich2/2.2 which is the only mvapich version on LC that grants
access to pnetcdf/1.9.0.

These changes reflect a change in the default module loads on LC
machines.

In addition, there is a change to the run_script to purge all mention
of the machines cab and sierra (which have been taken down) and to
add a warning message to the build section warning of the sensitivity
of CIME to double slashes (//) in the case directory path.

[BFB] - Bit-For-Bit

See confluence for a more detailed description about these tags.

When using pio2 if the pio_numiotasks and pio_stride indicate processors above the model task count you will get an error. This change prevents that happening.

Test suite: by hand SMS_P45.f19_g16.X
Test baseline:
Test namelist changes:
Test status: bit for bit
Fixes

User interface changes?:

Update gh-pages html (Y/N)?:

Code review:

…tory and finish refactoring dead nuopc components

…l of shr_sys_abort with ESMF_FAILURE tracing

Conflicts:
	src/share/nuopc/glc_elevclass_mod.F90

The call to cime_run_glc_setup_send was inadvertently deleted in
commit b24d24f

This commit reconciles changes in this branch with the contents of
cime_run_glc_setup_send.  Specifically, cime_run_glc_setup_send is
modified to separate steps needed for coupling from LND and OCN.  This
separation is also applied to prep_glc_accum_avg in prep_glc_mod.F90.
Finally, prep_glc_mrg and the routine it calls have been renamed to
indicate they are specific to forcing coming from LND.

Since the rof grid is identical to lnd, we should be able to use idmap
for the rof <-> lnd mappings.

* esmci_remote_for_split/master: (1573 commits)
  fix for unit test
  Fix handling of BGC and TEST compset elements. Also fix CESM config_files.xml
  bcast iac namelist settings
  Updates the assignment of oneletterid for components
  make oneletterid z
  generalize for BGC and TEST
  changes for TEST dataset and undefined lookups
  Adds CONFIG_IAC_FILE for CESM
  Adds missing XML entries in MCT
  Adds coupling interval and config file for IAC
  Fixes to grid and PIO
  need to USE_FMS to build mom
  Removes SIAC from compset definition
  Fix minor bugs
  Introduces a new stub IAC
  runtime fix
  removed med_connectors_mod.F90 and med_infodata_mod.F90
  fixed issues in PR
  Refactor code to make testable code a separate method
  Update for cime5.6.14
  ...

Upgrading the summit cmake, essl and MPI modules.

With the older version of MPI modules, MPI_Finalize call hangs.
The older version of essl module is no longer available.

Fixes #2847

…9-04-15

Jgfouca/branch for acme split 2019 04 15
Changes:

Improve robustness of wget and err msgs
Restore Aquaplanet Capability
Making sure that we print out PIO logs only from rank 0 of each component.
modified log message in shr_strdata_mod
Test suite: scripts-regression-tests
Test baseline:
Test namelist changes:
Test status: [bit for bit, roundoff, climate changing]

Fixes [CIME Github issue #]

User interface changes?:

Update gh-pages html (Y/N)?:

Code review: @jedwards4b

This PR adds first set of machine files for E3SM machine Compy. Machine
files for compilers Intel, gnu and PGI are added. I have used the
following test to compile and run the model:

SMS.ne4_ne4.FC5AV1C-L

For the above test:
-Intel compiler works fine
-GNU 4.8.5: Land model has a compile time error
-PGI: Netcdf is not built with this compiler yet

[BFB]

gnu compiler builds fine but blows up during run time with
following error:

set_time_float_from_date: error return from ESMF_TimeSet
for set_time_float_from_date
ERROR: CHKRC

[BFB] - Bit-For-Bit

Need consistent case for USE_CXX in Makefile. Intel should
not add qopenmp unless compile_threaded is TRUE.

A test has been modified to use a new AQUAP compset as the old
one doesn't exist anymore

[BFB] - Bit-For-Bit

Also add modules for intel+openmpi and reduce number of nodes
in default pelayouts for quicker testing.

Add NLDAS grid for CTSM and MOSART
Adds a high-resolution regional grid over the U.S. for use in land-only
simulations.

(I haven't added the necessary files to the inputdata repository yet,
but will do so once this and the corresponding CTSM PRs are accepted.)

Test suite: SMS test using this new grid in the context of a CTSM branch
with the necessary changes
Test baseline: n/a
Test namelist changes: none
Test status: bit for bit

Fixes none

User interface changes?: N

Update gh-pages html (Y/N)?: N

Code review:

This commits moves remaining details of the new OCN-GLC coupling from
the main cime_run() routine to subroutines.

A routine cime_run_ocnglc_coupling() does the actual calculations and is
called from cime_run_ocn_recv_post() to ensure it gets runs after each
time OCN completes.

This commit also creates separate prep_glc_accum_*() routines for
fields coming from LND and OCN so they can be called independently.
They are called from the respective cime_run_*_recv_post() routines for
hte component from which the fields are coming.

prep_glc_accum_avg() is still combined for fields coming from the two
components because this step happens in the same place for fields coming
from either LND or OCN.

Remove ESMF modules on edison

[BFB] - Bit-For-Bit

Adds machine files for Compy

This PR adds first set of machine files for E3SM machine Compy. Machine
files for compilers Intel, gnu and PGI are added. I have used the
following test to compile and run the model:

SMS.ne4_ne4.FC5AV1C-L

For the above test:
-Intel compiler works fine
-GNU 4.8.5: Land model has a compile time error
-PGI: Netcdf is not built using this compiler yet

[BFB]

The call from prep_ocn_mod.F90 is not needed.

Due to recent changes to cprnc, differences in field lists (i.e., if one
case has some diagnostic fields that are missing from the other case)
are now treated as differences. However, this is a problem for some new
tests that are wanted, such as ensuring that turning on carbon isotopes
doesn't change answers.

This commit allows a particular test type to dictate that field list
differences should be ignored. For tests that use SystemTestsCompareTwo,
this is as simple as adding an argument to the SystemTestsCompareTwo
constructor. For other tests, a new argument can be added to the call to
self._component_compare_test. (Differences in the values in shared
fields are still treated as differences.)

Allow specific tests to ignore field list differences
Due to recent changes to cprnc, differences in field lists (i.e., if one
case has some diagnostic fields that are missing from the other case)
are now treated as differences. However, this is a problem for some new
tests that are wanted, such as ensuring that turning on carbon isotopes
doesn't change answers.

This PR allows a particular test type to dictate that field list
differences should be ignored. For tests that use SystemTestsCompareTwo,
this is as simple as adding an argument to the SystemTestsCompareTwo
constructor. For other tests, a new argument can be added to the call to
self._component_compare_test. (Differences in the values in shared
fields are still treated as differences.)

Test suite: scripts_regression_tests on cheyenne; also manual testing as
noted below
Test baseline: n/a
Test namelist changes: none
Test status: bit for bit

Manual testing:

Ran a PEM test where I put in place a different copy of the '.base'
file as the test is running, which has an extra field: reported as
DIFFER, as expected.

Redid that test with this new flag: now reported as IDENTICAL; done
via the following diffs:

diff --git a/scripts/lib/CIME/SystemTests/pem.py b/scripts/lib/CIME/SystemTests/pem.py
index 34bb67a..abe2546 100644
--- a/scripts/lib/CIME/SystemTests/pem.py
+++ b/scripts/lib/CIME/SystemTests/pem.py
@@ -20,7 +20,8 @@ class PEM(SystemTestsCompareTwo):
                                        separate_builds = True,
                                        run_two_suffix = 'modpes',
                                        run_one_description = 'default pe counts',
-                                       run_two_description = 'halved pe counts')
+                                       run_two_description = 'halved pe counts',
+                                       ignore_fieldlist_diffs = True)

     def _case_one_setup(self):
         pass
If I instead put in place a file with diffs in one field along with
the added field, this is still reported as DIFFER (i.e., I didn't
break the reporting of true differences when adding
ignore_fieldlist_diffs)
Fixes: none

User interface changes?: N

Update gh-pages html (Y/N)?: N

Code review: @jedwards4b @jgfouca

Thanks to @klindsay28 for the inspiration for this change. Also cc @ekluzek .

…on mastertask

Adds the NETCDF_PATH to config_machines.xml for the Livermore
machines.  This is needed to sucessfully build cprnc, for
example.

[BFB] - Bit-For-Bit

See confluence for a more detailed description about these tags.

update the intel compiler for cheyenne

Fix ne0CONUSne30x8 grids
Update the ne0CONUSne30x8 mapping files, since the original ones used the wrong
SCRIP grid files.

Test suite: scripts_regression_tests
Test baseline:
Test namelist changes:
Test status: bit for bit

Fixes [CIME Github issue #]

User interface changes?:

Update gh-pages html (Y/N)?:

Code review:

OLCF recommends setting "OMPI_MCA_io" env variable to "romio314"
to prevent out of memory issues with the code. In our testing we
have found that setting this env gets rid of OOM errors with
certain E3SM simulations (ECP simulation runs).

However also note that setting this environment variable reduces
the performance of parallel HDF5 (when using NetCDF4P PIO iotype
to write data).

See Issue #2856

…e_fixes

Upgrading the summit cmake, essl and MPI modules.

Also updating the ROMIO version to prevent OOM errors.

Fixes #2847

[BFB]

Conflicts:
	src/drivers/nuopc/mediator/med_constants_mod.F90

Fix an issue in mct configure by providing the correct link flags for nag f90.

Test suite: SMS_D.f45_g37.A.izumi_nag
Test baseline:
Test namelist changes:
Test status: [bit for bit, roundoff, climate changing]

Fixes

User interface changes?:

Update gh-pages html (Y/N)?:

Code review:

…to srun via -c) for KNL.

[bfb]

Update CIME to ESMCI cime5.8.2

Squash merge of jgfouca/branch-for-to-acme-2019-04-22

Features:
* Adds a stub for Integrated Assessment Component (IAC)
* Allow specific tests to ignore field list differences

Bug fixes:
* Minor

[BFB]

Minor build fixes

Need consistent case for USE_CXX in Makefile (Restores cxxlib flag for
PIO2). Intel should not add qopenmp unless compile_threaded is TRUE.

Fixes #2858

[BFB]

* origin/jgfouca/cime/minor_compiler_fixes:
  Minor build fixes

scripts_regression_tests.py

Test suite:
Test baseline:
Test namelist changes:
Test status: roundoff

Fixes

User interface changes?:

Update gh-pages html (Y/N)?:

Code review:

Changes the config files in CIME to reflect recent changes in Livermore
Computing resources:

Update syrah and quartz machine config entries to explicitly load
mvapich2/2.2 which is the only mvapich version on LC that grants
access to pnetcdf/1.9.0.

These changes reflect a change in the default module loads on LC
machines.

In addition, there is a change to the run_script to purge all mention
of the machines cab and sierra (which have been taken down) and to
add a warning message to the build section warning of the sensitivity
of CIME to double slashes (//) in the case directory path.

[BFB] - Bit-For-Bit

This capability will be disabled for CESM.
This commit also corrects usage of glcshelf_c2_ocn instead of glc_c2_ocn in a couple places.

Whitespace, indentation, removal of unneeded comments, keyword arguments
for long argument lists.

In prep_glc_mod, there were two places where a list of coupling fields
from OCN were hardcoded in calls to seq_map_map.  This commit replaces
those hard-coded lists with a variable that is generated in seq_flds_mod
from the definitions of the coupling fields that are required.

These new coupling fields are disabled for CESM.

The OCN-GLC time averaging accumulators should only be read/written to
coupler restart files if OCN-GLC are actually coupled.  Otherwise there
will be an attempt to read/write unallocated things.

This required passing the ocn_c2_glc flag to the i/o routines, because
that information is not in infodata.

Nuopcnostub