Hi MOM6 enthusiasts:
I think I have seen the restart behavior documented online, but I'm afraid I cannot seem to locate it at the moment (googling "MOM6 restart how-to" "MOM6 restart configuration" does not locate it). Sorry if this is a newbie question.
Initially, I perform a short, 3-day run, which appears to work correctly (initially, the ./RESTART/ directory is empty). The following stanzas are in input.nml:
&MOM_input_nml
output_directory = 'OUTPUT',
input_filename = 'n'
restart_input_dir = 'RESTART/',
restart_output_dir = 'RESTART/',
parameter_filename = 'MOM_input',
'MOM_layout',
'MOM_override'
/
&ocean_solo_nml
calendar = 'gregorian',
date_init= 0,0,0,0,0,0,
months = 0,
days = 3,
hours = 0,
minutes = 6,
seconds = 0 /
Now, when I change the &ocean_solo_nml "days" parameter to 30, and I attempt to re-run it, I receive an error at runtime:
FATAL from PE 0: diag_time_inc: Invalid date. Date=0001-02-31 00:10:00
I can see that "0001-02-31" is indeed an invalid date: presumably in year 0001 of the Gregorian calendar there were not 31 days in February.
Unexpected behavior: When I deleted the restart files under ./RESTART/ the model ran correctly, without the date error.
Two questions:
(1) What is the recommended strategy for debugging the "Invalid date"?
I tried grepping through the source trees for the error message, and eventually I found the fragment "diag_time_inc" over in src/FMS/diag_manager/diag_util.F90. Should I compile with DEBUG=t to get more info from the call-stack?
(2) My recollection is that the 'n' parameter for &MOM_input_nml triggers a fresh run, and 'r' triggers a restart. Can you point me to the input.nml or restart documentation? According to this web page, Run-time Parameter System — MOM6 0.2a3 documentation, "input_filename- If equal to “n” will run a new run"
Additional information:
(1) This is all based on a recent checkout of MOM6-examples:
"git status" and "git log" says:
On branch dev/gfdl
Your branch is up to date with 'origin/dev/gfdl'
commit 433917f9ee529a3b331dcd2bb0d358f51256d6d6
Merge: 5bbcb803d 31a3475b7
Author: Robert Hallberg <Robert.Hallberg@noaa.gov>
Date: Fri Aug 2 09:37:50 2024 -0400
(2) The latest log from the main repo in my MOM6 source tree is:
commit b2db6bf6f7b76ba46c0e31cc61c075babea7c0fc
Author: Marshall Ward <marshall.ward@noaa.gov>
Date: Thu Sep 12 12:07:12 2024 -0400
Thank you for your attention.
-Ed
I think I have seen the restart behavior documented online, but I'm afraid I cannot seem to locate it at the moment (googling "MOM6 restart how-to" "MOM6 restart configuration" does not locate it). Sorry if this is a newbie question.
Initially, I perform a short, 3-day run, which appears to work correctly (initially, the ./RESTART/ directory is empty). The following stanzas are in input.nml:
&MOM_input_nml
output_directory = 'OUTPUT',
input_filename = 'n'
restart_input_dir = 'RESTART/',
restart_output_dir = 'RESTART/',
parameter_filename = 'MOM_input',
'MOM_layout',
'MOM_override'
/
&ocean_solo_nml
calendar = 'gregorian',
date_init= 0,0,0,0,0,0,
months = 0,
days = 3,
hours = 0,
minutes = 6,
seconds = 0 /
Now, when I change the &ocean_solo_nml "days" parameter to 30, and I attempt to re-run it, I receive an error at runtime:
FATAL from PE 0: diag_time_inc: Invalid date. Date=0001-02-31 00:10:00
I can see that "0001-02-31" is indeed an invalid date: presumably in year 0001 of the Gregorian calendar there were not 31 days in February.
Unexpected behavior: When I deleted the restart files under ./RESTART/ the model ran correctly, without the date error.
Two questions:
(1) What is the recommended strategy for debugging the "Invalid date"?
I tried grepping through the source trees for the error message, and eventually I found the fragment "diag_time_inc" over in src/FMS/diag_manager/diag_util.F90. Should I compile with DEBUG=t to get more info from the call-stack?
(2) My recollection is that the 'n' parameter for &MOM_input_nml triggers a fresh run, and 'r' triggers a restart. Can you point me to the input.nml or restart documentation? According to this web page, Run-time Parameter System — MOM6 0.2a3 documentation, "input_filename- If equal to “n” will run a new run"
Additional information:
(1) This is all based on a recent checkout of MOM6-examples:
"git status" and "git log" says:
On branch dev/gfdl
Your branch is up to date with 'origin/dev/gfdl'
commit 433917f9ee529a3b331dcd2bb0d358f51256d6d6
Merge: 5bbcb803d 31a3475b7
Author: Robert Hallberg <Robert.Hallberg@noaa.gov>
Date: Fri Aug 2 09:37:50 2024 -0400
(2) The latest log from the main repo in my MOM6 source tree is:
commit b2db6bf6f7b76ba46c0e31cc61c075babea7c0fc
Author: Marshall Ward <marshall.ward@noaa.gov>
Date: Thu Sep 12 12:07:12 2024 -0400
Thank you for your attention.
-Ed
