Use pregenerated (in the experiment generation step) values for ensem…

…ble member names and indices in those workflow config files in which this isn't already done; in some corresponding ex-scripts, generalize construction of ensemble member names to use ENSMEM_BASENAME and NDIGITS_IN_ENSMEM_NAMES; other related fixes.
gsketefian · Nov 28, 2024 · f327dcd · f327dcd
1 parent ca67bb6
commit f327dcd
Show file tree

Hide file tree

Showing 10 changed files with 64 additions and 55 deletions.
diff --git a/jobs/JREGIONAL_CHECK_POST_OUTPUT b/jobs/JREGIONAL_CHECK_POST_OUTPUT
@@ -8,7 +8,7 @@
 # Run-time environment variables:
 #
 #    CDATE
-#    ENSMEM_INDX
+#    ENSMEM_NAME
 #    GLOBAL_VAR_DEFNS_FP
 #
 # Experiment variables

diff --git a/parm/wflow/coldstart.yaml b/parm/wflow/coldstart.yaml
@@ -11,7 +11,7 @@ default_task: &default_task
     nprocs: '{{ parent.nnodes * parent.ppn }}'
     subcyc: !cycstr "@M"
     LOGDIR: !cycstr "&LOGDIR;"
-    ENSMEM_INDX: '#mem#'
+    ENSMEM_INDX: '#mem_indx#'
   native: '{{ platform.SCHED_NATIVE_CMD }}'
   nodes: '{{ nnodes }}:ppn={{ ppn }}'
   partition: '{% if platform.get("PARTITION_DEFAULT") %}&PARTITION_DEFAULT;{% else %}None{% endif %}'
@@ -82,8 +82,9 @@ task_get_extrn_lbcs:
 
 metatask_run_ensemble:
   var:
-    mem: '{% if global.DO_ENSEMBLE %}{%- for m in range(1, global.NUM_ENS_MEMBERS+1) -%}{{ "%03d "%m }}{%- endfor -%} {% else %}{{ "000"|string }}{% endif %}'
-  task_make_ics_mem#mem#:
+    mem_indx: '&ENSMEM_INDICES;'
+    mem_name: '&ENSMEM_NAMES;'
+  task_make_ics_#mem_name#:
     <<: *default_task
     command: '&LOAD_MODULES_RUN_TASK; "make_ics" "&JOBSdir;/JREGIONAL_MAKE_ICS"'
     envars:
@@ -122,7 +123,7 @@ metatask_run_ensemble:
             left: staged_sfc_climo
             right: '{% if not rocoto.get("tasks", {}).get("task_make_sfc_climo") %}staged_sfc_climo{% endif %}'
 
-  task_make_lbcs_mem#mem#:
+  task_make_lbcs_#mem_name#:
     <<: *default_task
     command: '&LOAD_MODULES_RUN_TASK; "make_lbcs" "&JOBSdir;/JREGIONAL_MAKE_LBCS"'
     envars:
@@ -140,7 +141,7 @@ metatask_run_ensemble:
           attrs:
             task: get_extrn_lbcs
 
-  task_run_fcst_mem#mem#:
+  task_run_fcst_#mem_name#:
     <<: *default_task
     command: '&LOAD_MODULES_RUN_TASK; "run_fcst" "&JOBSdir;/JREGIONAL_RUN_FCST"'
     envars:
@@ -157,10 +158,10 @@ metatask_run_ensemble:
       and:
         taskdep_make_ics:
           attrs:
-            task: make_ics_mem#mem#
+            task: make_ics_#mem_name#
         taskdep_make_lbcs:
           attrs:
-            task: make_lbcs_mem#mem#
+            task: make_lbcs_#mem_name#
         or_nexus_emission:
           not:
             taskvalid:

diff --git a/parm/wflow/default_workflow.yaml b/parm/wflow/default_workflow.yaml
@@ -27,6 +27,10 @@ rocoto:
     RUN: '{{ nco.RUN_default }}'
     SCRIPTSdir: '{{ user.SCRIPTSdir }}'
     SENDCOM: '{{ nco.SENDCOM_default }}'
+    SLASH_ENSMEM_SUBDIR: '{{- "/" }}
+                          {%- if global.DO_ENSEMBLE %}
+                            {{- "#mem_name#" }}
+                          {%- endif %}'
     ENSMEM_INDICES: '{%- for mem_indx in global.ENSMEM_INDS %}
                        {{- mem_indx ~ " " }}
                      {%- endfor %}'
@@ -43,7 +47,7 @@ rocoto:
     DATAROOT: '{{ nco.PTMP }}/&envir;/tmp'
     DCOMROOT: '{{ nco.PTMP }}/&envir;/dcom'
     COMIN_DIR: '{% if user.RUN_ENVIR == "nco" %}&COMROOT;/&NET;/&model_ver;/&RUN;.@Y@m@d/@H{% else %}{{ workflow.EXPTDIR }}/@Y@m@d@H{% endif %}'
-    FCST_DIR: '{% if user.RUN_ENVIR == "nco" %}&DATAROOT;/run_fcst_mem#mem#_@Y@m@d@H{% else %}{{ workflow.EXPTDIR }}/@Y@m@d@H{% endif %}'
+    FCST_DIR: '{% if user.RUN_ENVIR == "nco" %}&DATAROOT;/run_fcst_#mem_name#_@Y@m@d@H{% else %}{{ workflow.EXPTDIR }}/@Y@m@d@H{% endif %}'
     LOGDIR: '{% if user.RUN_ENVIR == "nco" %}&COMROOT;/output/logs/@Y@m@d{% else %}{{ workflow.EXPTDIR }}/log{% endif %}'
   attrs:
     cyclethrottle: "200"

diff --git a/parm/wflow/plot.yaml b/parm/wflow/plot.yaml
@@ -15,7 +15,7 @@ default_task_plot: &default_task
     fhr: '#fhr#'
     LOGDIR: !cycstr "&LOGDIR;"
     SLASH_ENSMEM_SUBDIR: '&SLASH_ENSMEM_SUBDIR;'
-    ENSMEM_INDX: '#mem#'
+    ENSMEM_INDX: '#mem_indx#'
     nprocs: '{{ parent.nnodes * parent.ppn }}'
   join: !cycstr '&LOGDIR;/{{ jobname }}_@Y@m@d@H&LOGEXT;'
   native: '{{ platform.SCHED_NATIVE_CMD }}'
@@ -28,29 +28,30 @@ default_task_plot: &default_task
 
 metatask_plot_allvars:
   var:
-    mem: '{% if global.DO_ENSEMBLE %}{%- for m in range(1, global.NUM_ENS_MEMBERS+1) -%}{{ "%03d "%m }}{%- endfor -%} {% else %}{{ "000"|string }}{% endif %}'
-  metatask_plot_allvars_mem#mem#_all_fhrs:
+    mem_indx: '&ENSMEM_INDICES;'
+    mem_name: '&ENSMEM_NAMES;'
+  metatask_plot_allvars_#mem_name#_all_fhrs:
     var:
       fhr: '{% for h in range(0, workflow.LONG_FCST_LEN+1) %}{{ " %03d" % h }}{% endfor %}'
       cycledef: '{% for h in range(0, workflow.LONG_FCST_LEN+1) %}{% if h <= workflow.FCST_LEN_CYCL|min %}forecast {% else %}long_forecast {% endif %}{% endfor %}'
-    task_plot_allvars_mem#mem#_f#fhr#:
+    task_plot_allvars_#mem_name#_f#fhr#:
       <<: *default_task
       command: '&LOAD_MODULES_RUN_TASK; "plot_allvars" "&JOBSdir;/JREGIONAL_PLOT_ALLVARS"'
       dependency:
         or_do_post: &post_files_exist
           and_run_post: # If post was meant to run, wait on the whole post metatask
             taskvalid:
               attrs:
-                task: run_post_mem#mem#_f#fhr#
+                task: run_post_#mem_name#_f#fhr#
             metataskdep:
               attrs:
                 metatask: run_ens_post
           and_inline_post: # If inline post ran, wait on the forecast task to complete
             not:
               taskvalid:
                 attrs:
-                  task: run_post_mem#mem#_f#fhr#
+                  task: run_post_#mem_name#_f#fhr#
             taskdep:
               attrs:
-                task: run_post_mem#mem#_f#fhr#
+                task: run_post_#mem_name#_f#fhr#
 
diff --git a/parm/wflow/post.yaml b/parm/wflow/post.yaml
@@ -11,7 +11,7 @@ default_task_post: &default_task
     cyc: !cycstr '@H'
     fhr: '#fhr#'
     SLASH_ENSMEM_SUBDIR: '&SLASH_ENSMEM_SUBDIR;'
-    ENSMEM_INDX: '#mem#'
+    ENSMEM_INDX: '#mem_indx#'
     nprocs: '{{ parent.nnodes * parent.ppn }}'
   join: !cycstr '&LOGDIR;/{{ jobname }}_@Y@m@d@H&LOGEXT;'
   native: '{{ platform.SCHED_NATIVE_CMD }}'
@@ -24,18 +24,19 @@ default_task_post: &default_task
 
 metatask_run_ens_post:
   var:
-    mem: '{% if global.DO_ENSEMBLE %}{%- for m in range(1, global.NUM_ENS_MEMBERS+1) -%}{{ "%03d "%m }}{%- endfor -%} {% else %}{{ "000"|string }}{% endif %}'
-  metatask_run_post_mem#mem#_all_fhrs:
+    mem_indx: '&ENSMEM_INDICES;'
+    mem_name: '&ENSMEM_NAMES;'
+  metatask_run_post_#mem_name#_all_fhrs:
     var:
       fhr: '{% for h in range(0, workflow.LONG_FCST_LEN+1) %}{{ " %03d" % h }}{% endfor %}'
       cycledef: '{% for h in range(0, workflow.LONG_FCST_LEN+1) %}{% if h <= workflow.FCST_LEN_CYCL|min %}forecast {% else %}long_forecast {% endif %}{% endfor %}'
-    task_run_post_mem#mem#_f#fhr#:
+    task_run_post_#mem_name#_f#fhr#:
       <<: *default_task
       dependency:
         or:
           taskdep:
             attrs:
-              task: run_fcst_mem#mem#
+              task: run_fcst_#mem_name#
           and:
             datadep_dyn:
               text: !cycstr '&FCST_DIR;&SLASH_ENSMEM_SUBDIR;/dynf#fhr#.nc'
@@ -52,7 +53,7 @@ metatask_run_ens_post:
     metatask_sub_hourly_post:
       var:
         fmn: '{% for min in range(0, 60, task_run_post.DT_SUBHOURLY_POST_MNTS) %}{{ " %02d" % min }}{% endfor %}'
-      task_run_post_mem#mem#_f#fhr##fmn#: &subhourly_run_post_task
+      task_run_post_#mem_name#_f#fhr##fmn#: &subhourly_run_post_task
         <<: *default_task
         envars:
           <<: *default_vars
@@ -61,7 +62,7 @@ metatask_run_ens_post:
           or:
             taskdep:
               attrs:
-                task: run_fcst_mem#mem#
+                task: run_fcst_#mem_name#
             and:
               datadep_dyn:
                 text: !cycstr '&FCST_DIR;&SLASH_ENSMEM_SUBDIR;/dynf#fhr#:#fmn#:00.nc'
@@ -76,5 +77,5 @@ metatask_run_ens_post:
     var:
       fhr: '{{ " %03d" % workflow.FCST_LEN_HRS }}'
       fmn: '00'
-    task_run_post_mem#mem#_f#fhr##fmn#:
+    task_run_post_#mem_name#_f#fhr##fmn#:
       <<: *subhourly_run_post_task
diff --git a/parm/wflow/prdgen.yaml b/parm/wflow/prdgen.yaml
@@ -1,11 +1,12 @@
 metatask_run_prdgen:
   var:
-    mem: '{% if global.DO_ENSEMBLE  %}{%- for m in range(1, global.NUM_ENS_MEMBERS+1) -%}{{ "%03d "%m }}{%- endfor -%} {% else %}{{ "000"|string }}{% endif %}'
-  metatask_run_prdgen_hrs_mem#mem#:
+    mem_indx: '&ENSMEM_INDICES;'
+    mem_name: '&ENSMEM_NAMES;'
+  metatask_run_prdgen_hrs_#mem_name#:
     var:
       fhr: '{% for h in range(0, workflow.LONG_FCST_LEN+1) %}{{ " %03d" % h }}{% endfor %}'
       cycledef: '{% for h in range(0, workflow.LONG_FCST_LEN+1) %}{% if h <= workflow.FCST_LEN_HRS %}forecast {% else %}long_forecast {% endif %}{% endfor %}'
-    task_run_prdgen_mem#mem#_f#fhr#:
+    task_run_prdgen_#mem_name#_f#fhr#:
       account: '&ACCOUNT;'
       attrs:
         cycledefs: '#cycledef#'
@@ -18,7 +19,7 @@ metatask_run_prdgen:
         cyc: !cycstr '@H'
         fhr: '#fhr#'
         SLASH_ENSMEM_SUBDIR: '&SLASH_ENSMEM_SUBDIR;'
-        ENSMEM_INDX: '#mem#'
+        ENSMEM_INDX: '#mem_indx#'
         nprocs: '{{ parent.nnodes * parent.ppn }}'
       join: !cycstr '&LOGDIR;/{{ jobname }}_@Y@m@d@H&LOGEXT;'
       native: '{{ platform.SCHED_NATIVE_CMD }}'
@@ -31,4 +32,4 @@ metatask_run_prdgen:
       dependency:
         taskdep:
           attrs:
-            task: run_post_mem#mem#_f#fhr#
+            task: run_post_#mem_name#_f#fhr#
diff --git a/parm/wflow/test.yaml b/parm/wflow/test.yaml
@@ -15,7 +15,7 @@ default_task_test: &default_task
     subcyc: !cycstr "@M"
     LOGDIR: !cycstr "&LOGDIR;"
     SLASH_ENSMEM_SUBDIR: '&SLASH_ENSMEM_SUBDIR;'
-    ENSMEM_INDX: '#mem#'
+    ENSMEM_INDX: '#mem_indx#'
   native: '{{ platform.SCHED_NATIVE_CMD }}'
   nnodes: 1
   nodes: '{{ nnodes }}:ppn={{ ppn }}'
@@ -26,17 +26,18 @@ default_task_test: &default_task
 
 metatask_integration_test:
   var:
-    mem: '{% if global.DO_ENSEMBLE  %}{%- for m in range(1, global.NUM_ENS_MEMBERS+1) -%}{{ "%03d "%m }}{%- endfor -%} {% else %}{{ "000"|string }}{% endif %}'
-  task_integration_test_mem#mem#:
+    mem_indx: '&ENSMEM_INDICES;'
+    mem_name: '&ENSMEM_NAMES;'
+  task_integration_test_#mem_name#:
     <<: *default_task
     command: '&LOAD_MODULES_RUN_TASK; "integration_test" "&JOBSdir;/JREGIONAL_INTEGRATION_TEST"'
     join: !cycstr '&LOGDIR;/{{ jobname }}_@Y@m@d@H&LOGEXT;'
     dependency:
       and_run_fcst: 
         taskvalid:
           attrs:
-            task: run_fcst_mem#mem#
+            task: run_fcst_#mem_name#
         taskdep:
           attrs:
-            task: run_fcst_mem#mem#
+            task: run_fcst_#mem_name#
 
diff --git a/parm/wflow/verify_pre.yaml b/parm/wflow/verify_pre.yaml
@@ -214,17 +214,14 @@ metatask_check_post_output_all_mems:
         # so, to ensure that the stand-alone post metatask for the current member
         # has completed for all forecast hours.
         and_standalone_post:
-          # This "or" is to check whether a stand-alane post task exists [either for
-          # mem000 (if running a deterministic forecast) or for mem001 (if running
-          # an ensemble forecast)].  Checking for forecast hour 000 is sufficient.  
-          # If so, it implies post is being run stand-alone.
-          or_post_task_exists: &post_task_exists
-            taskvalid_mem000:
-              attrs:
-                task: run_post_mem000_f000
-            taskvalid_mem001:
-              attrs:
-                task: run_post_mem001_f000
+          # This checks whether a stand-alane post task exists for the first ensemble
+          # member (which will be member 0 for a deterministic forecast or member 1
+          # for an ensemble forecast).  Checking for forecast hour 000 is sufficient.
+          # If so, it implies post is being run stand-alone (as opposed to inline,
+          # i.e. from within the weather model).
+          taskvalid_first_ensmem: &post_task_exists
+            attrs:
+              task: '{{- "run_post_" ~ global.ENSMEM_NAMES[0] ~ "_f000" }}'
           # We would like a dependency here on the run_post_#mem_name#_all_fhrs metatask,
           # but that is apparently not possible in rocoto with a metataskvalid/
           # metataskdep combination (currently, metataskvalid is not a defined
@@ -254,13 +251,13 @@ metatask_check_post_output_all_mems:
         # the weather model), and if so, to ensure that the forecast task for the
         # current member has completed.
         and_inline_post:
-          # This "not" is to check whether a stand-alone post task DOESN'T exist
-          # [either for mem000 (if running a deterministic forecast) or for mem001
-          # (if running an ensemble forecast)].  If so, it implies that either
-          # post is being run inline (as part of the weather model) or is not being
-          # run at all.
+          # This "not" is to check whether a stand-alone post task DOESN'T exist.
+          # If it does, it will either be for ensemble member 0 for a deterministic
+          # forecast or for member 1 for an ensemble forecast.  If such a post task
+          # doesn't exist, it implies that either post is being run inline (i.e. as
+          # part of the weather model) or is not being run at all.
           not: 
-            or_post_task_exists:
+            taskvalid_first_ensmem:
               <<: *post_task_exists
           # This "or" is to verify that either the forecast task for the current
           # member exists and has completed, or that it doesn't exist.

diff --git a/scripts/exregional_get_extrn_mdl_files.sh b/scripts/exregional_get_extrn_mdl_files.sh
@@ -56,6 +56,8 @@
 #  global:
 #    DO_ENSEMBLE
 #    NUM_ENS_MEMBERS
+#    ENSMEM_BASENAME
+#    NDIGITS_IN_ENSMEM_NAMES
 #
 #-----------------------------------------------------------------------
 #
@@ -217,7 +219,7 @@ if [ $(boolify $SYMLINK_FIX_FILES) = "TRUE" ]; then
 fi
 
 if [ $(boolify $DO_ENSEMBLE) = "TRUE" ] ; then
-  mem_dir="/mem{mem:03d}"
+  mem_dir="/${ENSMEM_BASENAME}{mem:0${NDIGITS_IN_ENSMEM_NAMES}d}"
   member_list=(1 ${NUM_ENS_MEMBERS})
   additional_flags="$additional_flags \
   --members ${member_list[@]}"
@@ -287,7 +289,8 @@ if [ "${EXTRN_MDL_NAME}" = "GEFS" ]; then
         sorted_fn=( )
         for fcst_hr in "${all_fcst_hrs_array[@]}"; do
             # Read in filenames from EXTRN_MDL_FNS and sort them
-            base_path="${EXTRN_MDL_STAGING_DIR}/mem`printf %03d $num`"
+            ensmem_indx_fmt=$(printf "%0${NDIGITS_IN_ENSMEM_NAMES}d" $((10#${num})))
+            base_path="${EXTRN_MDL_STAGING_DIR}/${ENSMEM_BASENAME}${ensmem_indx_fmt}"
             filenames_array=`awk -F= '/EXTRN_MDL_FNS/{print $2}' $base_path/${EXTRN_DEFNS}`
             for filename in ${filenames_array[@]}; do
                 IFS='.' read -ra split_fn <<< "$filename"

diff --git a/scripts/exregional_run_post.sh b/scripts/exregional_run_post.sh
@@ -11,7 +11,7 @@
 #    COMOUT
 #    DATA_FHR
 #    DBNROOT
-#    ENSMEM_INDX
+#    ENSMEM_NAME
 #    GLOBAL_VAR_DEFNS_FP
 #    NET
 #    PDY
@@ -226,7 +226,7 @@ fi
 # Set the names of the forecast model's write-component output files.
 #
 if [ "${RUN_ENVIR}" = "nco" ]; then
-    DATAFCST=$DATAROOT/run_fcst_mem${ENSMEM_INDX}.${share_pid}
+    DATAFCST=$DATAROOT/run_fcst_${ENSMEM_NAME}.${share_pid}
 else
     DATAFCST=$DATA
 fi