From 3d0f42768303e91e9677f911567be43f78e74e0a Mon Sep 17 00:00:00 2001
From: DavidHuber-NOAA <david.huber@noaa.gov>
Date: Thu, 20 Nov 2025 17:34:13 +0000
Subject: [PATCH 01/35] Enable set -o pipefail to catch errors in piped
 commands

---
 ush/preamble.sh | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/ush/preamble.sh b/ush/preamble.sh
index fab7d163a93..439ad0d6bea 100755
--- a/ush/preamble.sh
+++ b/ush/preamble.sh
@@ -33,8 +33,8 @@ declare -x PS4='+ $(basename ${BASH_SOURCE[0]:-${FUNCNAME[0]:-"Unknown"}})[${LIN
 
 set_strict() {
     if [[ ${STRICT:-"YES"} == "YES" ]]; then
-        # Exit on error and undefined variable
-        set -eu
+        # Exit on error, undefined variable, or error in a pipeline (e.g. if and command in "cmd | cmd2" fails)
+        set -euo pipefail
     fi
 }
 

From 4ad6542239d1c48008ba9882a07da1d1b3f40fef Mon Sep 17 00:00:00 2001
From: DavidHuber-NOAA <david.huber@noaa.gov>
Date: Thu, 20 Nov 2025 17:35:07 +0000
Subject: [PATCH 02/35] Refactor cleanup task and add checks for gempak files

---
 dev/parm/config/gfs/config.cleanup |  24 ++++---
 scripts/exglobal_cleanup.sh        | 100 +++++++++++++++++++----------
 2 files changed, 82 insertions(+), 42 deletions(-)

diff --git a/dev/parm/config/gfs/config.cleanup b/dev/parm/config/gfs/config.cleanup
index 543d624caf2..62b4aca2efd 100644
--- a/dev/parm/config/gfs/config.cleanup
+++ b/dev/parm/config/gfs/config.cleanup
@@ -9,22 +9,30 @@ source "${EXPDIR}/config.resources" cleanup
 export CLEANUP_COM="YES"   # NO=retain ROTDIR.  YES default in cleanup.sh
 
 #--starting and ending hours of previous cycles to be removed from rotating directory
-export RMOLDSTD=144
-export RMOLDEND=24
+# Selectively remove files between SELECTIVE_RM_MIN and SELECTIVE_RM_MIN hours old, based on exclude_string
+# Remove all RTOFS files older than RTOFS_CLEANUP_MAX hours
+# Remove all files, except GEMPAK files, older than SELECTIVE_RM_MAX hours, based on gempak_exclude_string
+# Remove all files older than the max of all SELECTIVE_RM* variables
+# Retain all files newer than SELECTIVE_RM_MIN hours
 
-if [[ "${DO_GEMPAK}" == "YES" && "${RUN}" == "gfs" ]]; then
-    export RMOLDSTD=346
-    export RMOLDEND=222
+export SELECTIVE_CLEANUP_MAX=144
+export SELECTIVE_CLEANUP_MIN=24
+export GEMPAK_CLEANUP_MAX=240
+export RTOFS_CLEANUP_MAX=48
+if [[ ${SELECTIVE_CLEANUP_MIN} -gt ${SELECTIVE_CLEANUP_MAX} ]]; then
+    echo "FATAL ERROR: Invalid selective cleanup times: "
+    echo "             SELECTIVE_CLEANUP_MIN=${SELECTIVE_CLEANUP_MIN} > SELECTIVE_CLEANUP_MAX=${SELECTIVE_CLEANUP_MAX}"
+    exit 1
 fi
 
 # Specify the list of files to exclude from the first stage of cleanup
 # Because arrays cannot be exported, list is a single string of comma-
 # separated values. This string is split to form an array at runtime.
+export exclude_string=""
 case ${RUN} in
-	gdas | gfs)	exclude_string="*prepbufr*, *cnvstat.tar*, *analysis.atm.a*.nc" ;;
-	enkf*)      exclude_string="*f006.ens*" ;;
+	gdas | gfs)	exclude_string+="*prepbufr*, *cnvstat.tar*, *analysis.atm.a*.nc" ;;
+	enkf*)      exclude_string+="*f006.ens*" ;;
 	*)			exclude_string="" ;;
 esac
-export exclude_string
 
 echo "END: config.cleanup"
diff --git a/scripts/exglobal_cleanup.sh b/scripts/exglobal_cleanup.sh
index c460cc8282c..b6036b97211 100755
--- a/scripts/exglobal_cleanup.sh
+++ b/scripts/exglobal_cleanup.sh
@@ -6,7 +6,9 @@ echo "Begin Cleanup ${DATAROOT}!"
 # Remove DATAoutput from the forecast model run
 # TODO: Handle this better
 DATAfcst="${DATAROOT}/${RUN}fcst.${PDY:-}${cyc}"
-if [[ -d "${DATAfcst}" ]]; then rm -rf "${DATAfcst}"; fi
+if [[ -d "${DATAfcst}" ]];
+    then rm -rf "${DATAfcst}";
+fi
 #DATAefcs="${DATAROOT}/${RUN}efcs???${PDY:-}${cyc}"
 rm -rf "${DATAROOT}/${RUN}efcs"*"${PDY:-}${cyc}"
 ###############################################################
@@ -15,15 +17,32 @@ if [[ "${CLEANUP_COM:-YES}" == NO ]] ; then
     exit 0
 fi
 
+SELECTIVE_CLEANUP_MIN=${SELECTIVE_CLEANUP_MIN:-24}
+SELECTIVE_CLEANUP_MAX=${SELECTIVE_CLEANUP_MAX:-120}
+RTOFS_CLEANUP_MAX=${RTOFS_CLEANUP_MAX:-48}
+GEMPAK_CLEANUP_MAX=${GEMPAK_CLEANUP_MAX:-240}
 ###############################################################
 # Clean up previous cycles; various depths
 
 # Step back every assim_freq hours and remove old rotating directories
 # for successful cycles (defaults from 24h to 120h).
 # Retain files needed by Fit2Obs
-last_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${RMOLDEND:-24} hours")
-first_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${RMOLDSTD:-120} hours")
-last_rtofs=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${RMOLDRTOFS:-48} hours")
+last_selective_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${SELECTIVE_CLEANUP_MIN} hours")
+first_selective_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${SELECTIVE_CLEANUP_MAX} hours")
+last_rtofs_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${RTOFS_CLEANUP_MAX} hours")
+last_gempak_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${GEMPAK_CLEANUP_MAX} hours")
+exclude_string="${exclude_string:-}"
+
+# Find the last date among all cleanup targets
+max_cleanup_max="${SELECTIVE_CLEANUP_MAX:-120}"
+for cleanup_max in "${RTOFS_CLEANUP_MAX}" "${GEMPAK_CLEANUP_MAX}"; do
+    if [[ ${cleanup_max} -gt ${max_cleanup_max} ]]; then
+        max_cleanup_max=${cleanup_max}
+    fi
+done
+
+last_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${max_cleanup_max} hours")
+
 function remove_files() {
     local directory=$1
     shift
@@ -31,53 +50,63 @@ function remove_files() {
         echo "No directory ${directory} to remove files from, skiping"
         return
     fi
-    local find_exclude_string=""
-    for exclude in "$@"; do
-        find_exclude_string+="${find_exclude_string} -name ${exclude} -or"
+    # Find all files and links in the directory and store as an arry
+    # Run find only once for efficiency
+    flist=($(find "${directory}" -type f -or -type l))
+
+    # Now remove those files that match the exclude patterns
+    for exclude_pattern in "$@"; do
+        # Use a temporary array to hold files that do not match the exclude pattern
+        temp_flist=()
+        for file in "${flist[@]}"; do
+            case "$(basename "${file}")" in
+                ${exclude_pattern})
+                    # Match found, skip this file
+                    ;;
+                *)
+                    # No match, keep this file
+                    temp_flist+=("${file}")
+                    ;;
+            esac
+        done
+        flist=("${temp_flist[@]}")
     done
-    # Chop off any trailing or
-    find_exclude_string="${find_exclude_string[*]/%-or}"
-    # Remove all regular files that do not match
-    # shellcheck disable=SC2086
-    if [[ -n "${find_exclude_string}" ]]; then
-    # String is non-empty → use exclusion
-       find "${directory}" -type f -not \( ${find_exclude_string} \) -ignore_readdir_race -delete
-    else
-    # String is empty → no exclusion
-       find "${directory}" -type f -ignore_readdir_race -delete
-    fi
 
-    # Remove all symlinks that do not match
-    # shellcheck disable=SC2086
-    if [[ -n "${find_exclude_string}" ]]; then
-    # String is non-empty → use exclusion
-       find "${directory}" -type l -not \( ${find_exclude_string} \) -ignore_readdir_race -delete
-    else
-    # String is empty → no exclusion
-       find "${directory}" -type l -ignore_readdir_race -delete
-    fi
+    # Delete all files in flist.
+    for file in "${flist[@]}"; do
+        rm -f "${file}"
+    done
 
     # Remove any empty directories
     find "${directory}" -type d -empty -delete
 }
 
-for (( current_date=first_date; current_date <= last_date; \
+# Now start removing old COM files/directories
+for (( current_date=first_selective_date; current_date <= last_date; \
   current_date=$(date --utc +%Y%m%d%H -d "${current_date:0:8} ${current_date:8:2} +${assim_freq} hours") )); do
     current_PDY="${current_date:0:8}"
     current_cyc="${current_date:8:2}"
-    rtofs_dir="${ROTDIR}/rtofs.${current_PDY}"
     rocotolog="${EXPDIR}/logs/${current_date}.log"
+
+    # Extend the exclude list for gempak files if needed
+    if [[ "${RUN}" == "gfs" && ${current_date} -lt ${last_gempak_date} && "${DO_GEMPAK}" == "YES" ]]; then
+        # Provide the gempak exclude pattern(s)
+        exclude_string+=", *gfs_1p00_*"
+    fi
+
+    # Check if the cycle completed successfully by looking at the rocoto log
     if [[ -f "${rocotolog}" ]]; then
         # TODO: This needs to be revamped to not look at the rocoto log.
-        # shellcheck disable=SC2312
         if [[ $(tail -n 1 "${rocotolog}") =~ "This cycle is complete: Success" ]]; then
             YMD="${current_PDY}" HH="${current_cyc}" declare_from_tmpl \
                 COMOUT_TOP:COM_TOP_TMPL
             if [[ -d "${COMOUT_TOP}" ]]; then
-                IFS=", " read -r -a exclude_list <<< "${exclude_string:-}"
+                IFS=", " read -r -a exclude_list <<< "${exclude_string}"
                 remove_files "${COMOUT_TOP}" "${exclude_list[@]:-}"
             fi
-            if [[ -d "${rtofs_dir}" ]] && (( current_date < last_rtofs )); then rm -rf "${rtofs_dir}" ; fi
+            # Remove all rtofs directories in each RUN older than last_rtofs_date
+            rtofs_dir="${ROTDIR}/rtofs.${current_PDY}"
+            if [[ -d "${rtofs_dir}" ]] && (( current_date < last_rtofs_date )); then rm -rf "${rtofs_dir}" ; fi
         fi
     fi
 done
@@ -108,13 +137,16 @@ if [[ "${RUN}" == "gfs" ]]; then
 fi
 
 # Remove $RUN.$rPDY for the older of GDATE or RDATE
-GDATE=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${RMOLDSTD:-120} hours")
+GDATE=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${max_cleanup_max} hours")
 RDATE=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${FHMAX_GFS} hours")
 if (( GDATE < RDATE )); then
     RDATE=${GDATE}
 fi
+
 deletion_target="${ROTDIR}/${RUN}.${RDATE:0:8}"
-if [[ -d ${deletion_target} ]]; then rm -rf "${deletion_target}"; fi
+if [[ -d "${deletion_target}" ]]; then
+    rm -rf "${deletion_target}"
+fi
 
 # sync and wait to avoid filesystem synchronization issues
 sync && sleep 1

From 8e4322b572e7ffe30c0b97c5fabbe001e15c7def Mon Sep 17 00:00:00 2001
From: DavidHuber-NOAA <david.huber@noaa.gov>
Date: Thu, 20 Nov 2025 17:52:02 +0000
Subject: [PATCH 03/35] Shellcheck fixes

---
 scripts/exglobal_cleanup.sh | 15 +++++----------
 1 file changed, 5 insertions(+), 10 deletions(-)

diff --git a/scripts/exglobal_cleanup.sh b/scripts/exglobal_cleanup.sh
index b6036b97211..901d646f05f 100755
--- a/scripts/exglobal_cleanup.sh
+++ b/scripts/exglobal_cleanup.sh
@@ -52,22 +52,17 @@ function remove_files() {
     fi
     # Find all files and links in the directory and store as an arry
     # Run find only once for efficiency
-    flist=($(find "${directory}" -type f -or -type l))
+    mapfile -t flist < <(find "${directory}" -type f -or -type l)
 
     # Now remove those files that match the exclude patterns
     for exclude_pattern in "$@"; do
         # Use a temporary array to hold files that do not match the exclude pattern
         temp_flist=()
         for file in "${flist[@]}"; do
-            case "$(basename "${file}")" in
-                ${exclude_pattern})
-                    # Match found, skip this file
-                    ;;
-                *)
-                    # No match, keep this file
-                    temp_flist+=("${file}")
-                    ;;
-            esac
+            # shellcheck disable=SC2254
+            if [[ ! $(basename "${file}") == ${exclude_pattern} ]]; then
+                temp_flist+=("${file}")
+            fi
         done
         flist=("${temp_flist[@]}")
     done

From c6a23cfec21eb0d5e10fe48b55bafb307f8a3526 Mon Sep 17 00:00:00 2001
From: DavidHuber-NOAA <david.huber@noaa.gov>
Date: Thu, 20 Nov 2025 20:56:18 +0000
Subject: [PATCH 04/35] Ignore 2312

---
 .shellcheckrc               | 3 +++
 scripts/exglobal_cleanup.sh | 2 +-
 2 files changed, 4 insertions(+), 1 deletion(-)

diff --git a/.shellcheckrc b/.shellcheckrc
index 67fabfe1576..11ecc905ddf 100644
--- a/.shellcheckrc
+++ b/.shellcheckrc
@@ -17,3 +17,6 @@ disable=SC2174
 
 # Disable warning of functions in test statements
 disable=SC2310
+
+# Disable warning of piped commands masking exit codes; this is covered by set -o pipefail
+disable=SC2312
diff --git a/scripts/exglobal_cleanup.sh b/scripts/exglobal_cleanup.sh
index 901d646f05f..fb51627185f 100755
--- a/scripts/exglobal_cleanup.sh
+++ b/scripts/exglobal_cleanup.sh
@@ -59,7 +59,7 @@ function remove_files() {
         # Use a temporary array to hold files that do not match the exclude pattern
         temp_flist=()
         for file in "${flist[@]}"; do
-            # shellcheck disable=SC2254
+            # shellcheck disable=SC2053
             if [[ ! $(basename "${file}") == ${exclude_pattern} ]]; then
                 temp_flist+=("${file}")
             fi

From 484235c9bd1d5c2b63ee93d6b9fc8917fc67ece5 Mon Sep 17 00:00:00 2001
From: DavidHuber-NOAA <david.huber@noaa.gov>
Date: Thu, 20 Nov 2025 20:58:13 +0000
Subject: [PATCH 05/35] Remove unused variable

---
 scripts/exglobal_cleanup.sh | 1 -
 1 file changed, 1 deletion(-)

diff --git a/scripts/exglobal_cleanup.sh b/scripts/exglobal_cleanup.sh
index fb51627185f..8f4c71086f6 100755
--- a/scripts/exglobal_cleanup.sh
+++ b/scripts/exglobal_cleanup.sh
@@ -27,7 +27,6 @@ GEMPAK_CLEANUP_MAX=${GEMPAK_CLEANUP_MAX:-240}
 # Step back every assim_freq hours and remove old rotating directories
 # for successful cycles (defaults from 24h to 120h).
 # Retain files needed by Fit2Obs
-last_selective_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${SELECTIVE_CLEANUP_MIN} hours")
 first_selective_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${SELECTIVE_CLEANUP_MAX} hours")
 last_rtofs_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${RTOFS_CLEANUP_MAX} hours")
 last_gempak_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${GEMPAK_CLEANUP_MAX} hours")

From 749f91737f028497031af9285218ae79d60f29e6 Mon Sep 17 00:00:00 2001
From: David Huber <david.huber@noaa.gov>
Date: Fri, 21 Nov 2025 14:28:50 +0000
Subject: [PATCH 06/35] Disable pipefail exits for now

---
 .shellcheckrc   | 3 ---
 ush/preamble.sh | 5 +++--
 2 files changed, 3 insertions(+), 5 deletions(-)

diff --git a/.shellcheckrc b/.shellcheckrc
index 11ecc905ddf..67fabfe1576 100644
--- a/.shellcheckrc
+++ b/.shellcheckrc
@@ -17,6 +17,3 @@ disable=SC2174
 
 # Disable warning of functions in test statements
 disable=SC2310
-
-# Disable warning of piped commands masking exit codes; this is covered by set -o pipefail
-disable=SC2312
diff --git a/ush/preamble.sh b/ush/preamble.sh
index 439ad0d6bea..2b418c58173 100755
--- a/ush/preamble.sh
+++ b/ush/preamble.sh
@@ -33,8 +33,9 @@ declare -x PS4='+ $(basename ${BASH_SOURCE[0]:-${FUNCNAME[0]:-"Unknown"}})[${LIN
 
 set_strict() {
     if [[ ${STRICT:-"YES"} == "YES" ]]; then
-        # Exit on error, undefined variable, or error in a pipeline (e.g. if and command in "cmd | cmd2" fails)
-        set -euo pipefail
+        # Exit on error or undefined variable
+        # TODO: Also error in a pipeline (e.g. if and command in "cmd | cmd2" fails)
+        set -eu # -o pipefail
     fi
 }
 

From ceeae5e4713ee15337724a04afbe05a3aa065822 Mon Sep 17 00:00:00 2001
From: DavidHuber-NOAA <david.huber@noaa.gov>
Date: Fri, 21 Nov 2025 14:41:20 +0000
Subject: [PATCH 07/35] Shellcheck issues

---
 scripts/exglobal_cleanup.sh | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/scripts/exglobal_cleanup.sh b/scripts/exglobal_cleanup.sh
index 8f4c71086f6..13dbc5beb32 100755
--- a/scripts/exglobal_cleanup.sh
+++ b/scripts/exglobal_cleanup.sh
@@ -51,7 +51,8 @@ function remove_files() {
     fi
     # Find all files and links in the directory and store as an arry
     # Run find only once for efficiency
-    mapfile -t flist < <(find "${directory}" -type f -or -type l)
+    flist=()
+    mapfile -t flist < <(find "${directory}" -type f -or -type l) || true
 
     # Now remove those files that match the exclude patterns
     for exclude_pattern in "$@"; do
@@ -91,7 +92,9 @@ for (( current_date=first_selective_date; current_date <= last_date; \
     # Check if the cycle completed successfully by looking at the rocoto log
     if [[ -f "${rocotolog}" ]]; then
         # TODO: This needs to be revamped to not look at the rocoto log.
-        if [[ $(tail -n 1 "${rocotolog}") =~ "This cycle is complete: Success" ]]; then
+        tail_log=$(tail -n 1 "${rocotolog}") || true
+        # Test if the last line of rocotolog indicates success
+        if [[ ${tail_log} =~ "This cycle is complete: Success" ]]; then
             YMD="${current_PDY}" HH="${current_cyc}" declare_from_tmpl \
                 COMOUT_TOP:COM_TOP_TMPL
             if [[ -d "${COMOUT_TOP}" ]]; then

From 7006ec5af54d604a023355e50889cbdd6bc5047f Mon Sep 17 00:00:00 2001
From: DavidHuber-NOAA <david.huber@noaa.gov>
Date: Fri, 21 Nov 2025 14:52:40 +0000
Subject: [PATCH 08/35] Fix bufrsnd tarball creation

---
 scripts/exgfs_atmos_postsnd.sh | 11 +++++------
 1 file changed, 5 insertions(+), 6 deletions(-)

diff --git a/scripts/exgfs_atmos_postsnd.sh b/scripts/exgfs_atmos_postsnd.sh
index a366b1fad50..24408b872f5 100755
--- a/scripts/exgfs_atmos_postsnd.sh
+++ b/scripts/exgfs_atmos_postsnd.sh
@@ -148,12 +148,11 @@ else
 fi
 ${runscript} "${fhr}" "${fhr_p}" "${FINT}" "${F00FLAG}" "${DATA}"
 
-##############################################################
-# Tar and gzip the individual bufr files and send them to /com
-##############################################################
-cd "${COMOUT_ATMOS_BUFR}" || exit 2
-tar -cf - . | /usr/bin/gzip > "${RUN}.${cycle}.bufrsnd.tar.gz"
-cd "${DATA}" || exit 2
+###################################################
+# Tar and gzip the bufr files created placed in COM
+###################################################
+tar -czf "${RUN}.${cycle}.bufrsnd.tar.gz" -C "${COMOUT_ATMOS_BUFR}" bufr.*
+cpfs "${RUN}.${cycle}.bufrsnd.tar.gz" "${COMOUT_ATMOS_BUFR}/"
 
 ########################################
 # Send the single tar file to OSO

From 4efde5ada950c7bb39658f32f7fdd1b97cfc36ca Mon Sep 17 00:00:00 2001
From: DavidHuber-NOAA <david.huber@noaa.gov>
Date: Fri, 21 Nov 2025 19:18:44 +0000
Subject: [PATCH 09/35] Correct archiving to actually pull the data used

---
 jobs/JGFS_ATMOS_POSTSND        |  1 +
 scripts/exgfs_atmos_postsnd.sh | 35 ++++++++++++++++++++--------------
 ush/gfs_sndp.sh                | 14 ++++++++------
 3 files changed, 30 insertions(+), 20 deletions(-)

diff --git a/jobs/JGFS_ATMOS_POSTSND b/jobs/JGFS_ATMOS_POSTSND
index d2d7491f9fb..10f6248081e 100755
--- a/jobs/JGFS_ATMOS_POSTSND
+++ b/jobs/JGFS_ATMOS_POSTSND
@@ -12,6 +12,7 @@ source "${HOMEgfs}/ush/jjob_header.sh" -e "postsnd" -c "base postsnd"
 
 YMD=${PDY} HH=${cyc} declare_from_tmpl -rx \
     COMIN_ATMOS_HISTORY:COM_ATMOS_HISTORY_TMPL \
+    COMIN_ATMOS_BUFR:COM_ATMOS_BUFR_TMPL \
     COMOUT_ATMOS_BUFR:COM_ATMOS_BUFR_TMPL \
     COMOUT_ATMOS_WMO:COM_ATMOS_WMO_TMPL \
     COMOUT_ATMOS_GEMPAK:COM_ATMOS_GEMPAK_TMPL
diff --git a/scripts/exgfs_atmos_postsnd.sh b/scripts/exgfs_atmos_postsnd.sh
index 24408b872f5..559c074a0fa 100755
--- a/scripts/exgfs_atmos_postsnd.sh
+++ b/scripts/exgfs_atmos_postsnd.sh
@@ -148,20 +148,6 @@ else
 fi
 ${runscript} "${fhr}" "${fhr_p}" "${FINT}" "${F00FLAG}" "${DATA}"
 
-###################################################
-# Tar and gzip the bufr files created placed in COM
-###################################################
-tar -czf "${RUN}.${cycle}.bufrsnd.tar.gz" -C "${COMOUT_ATMOS_BUFR}" bufr.*
-cpfs "${RUN}.${cycle}.bufrsnd.tar.gz" "${COMOUT_ATMOS_BUFR}/"
-
-########################################
-# Send the single tar file to OSO
-########################################
-if [[ "${SENDDBN}" == "YES" ]]; then
-    "${DBNROOT}/bin/dbn_alert" MODEL GFS_BUFRSND_TAR "${job}" \
-        "${COMOUT_ATMOS_BUFR}/${RUN}.${cycle}.bufrsnd.tar.gz"
-fi
-
 ########################################
 # Create Regional Collectives of BUFR
 # data and add appropriate WMO Headers
@@ -182,6 +168,27 @@ chmod +x cmdfile
 
 ${APRUN_POSTSNDCFP} cmdfile
 
+###############################################################
+# Tar and gzip the bufr files in used to create the collectives
+###############################################################
+# The lists of files are located in ${DATA}/<m>/bufr_snd_files.list
+for (( m = 1; m <= NUM_SND_COLLECTIVES; m++ )); do
+   cat "${DATA}/${m}/bufr_snd_files.list" >> "${DATA}/all_bufr_snd_files.list"
+done
+
+tar -czf "${RUN}.${cycle}.bufrsnd.tar.gz" -T "${DATA}/all_bufr_snd_files.list" -C "${COMIN_ATMOS_BUFR}"
+
+cpfs "${RUN}.${cycle}.bufrsnd.tar.gz" "${COMOUT_ATMOS_BUFR}/"
+
+########################################
+# Send the single tar file to OSO
+########################################
+if [[ "${SENDDBN}" == "YES" ]]; then
+    "${DBNROOT}/bin/dbn_alert" MODEL GFS_BUFRSND_TAR "${job}" \
+        "${COMOUT_ATMOS_BUFR}/${RUN}.${cycle}.bufrsnd.tar.gz"
+fi
+
+
 ########################################
 # Read BUFR output and transfer into
 # GEMPAK surface and sounding data files
diff --git a/ush/gfs_sndp.sh b/ush/gfs_sndp.sh
index 3d1cd01401c..703e73845fa 100755
--- a/ush/gfs_sndp.sh
+++ b/ush/gfs_sndp.sh
@@ -10,7 +10,7 @@
   #  Create "collectives" consisting of groupings of the soundings
   #  into files designated by geographical region.   Each input
   #  file gfs_collective*.list (1-9) contains the list of stations to
-  #  put in a particular collective output file. 
+  #  put in a particular collective output file.
 export m=$1
 mkdir $DATA/$m
 cd $DATA/$m
@@ -19,18 +19,21 @@ cd $DATA/$m
     file_list=gfs_collective${m}.list
 
     if [ $m -le 2 ]
-    then 
+    then
       WMOHEAD=JUSA4$m
     elif [ $m -le 6 ]
-    then 
+    then
       WMOHEAD=JUSB4$m
-    else 
+    else
       WMOHEAD=JUSX4$m
     fi
 
     for stn in $(cat $file_list)
     do
-       cpreq "${COMOUT_ATMOS_BUFR}/bufr.${stn}.${PDY}${cyc}" "${DATA}/${m}/bufrin"
+       bufr_target="bufr.${stn}.${PDY}${cyc}"
+       cpreq "${COMIN_ATMOS_BUFR}/${bufr_target}" "${DATA}/${m}/bufrin"
+       # Also write the filename to a list for files to tar
+       echo "${bufr_target}" >> "${DATA}/${m}/bufr_snd_files.list"
        export pgm=tocsbufr.x
        #. prep_step
        export FORT11=$DATA/${m}/bufrin
@@ -59,4 +62,3 @@ EOF
 				   "${COMOUT_ATMOS_WMO}/gfs_collective${m}.postsnd_${cyc}"
     fi
     cpfs "${DATA}/${m}/gfs_collective${m}.fil" "${COMOUT_ATMOS_BUFR}/."
-

From 214eb5056aa7321ea53ad2297dccbcaf51158ff0 Mon Sep 17 00:00:00 2001
From: DavidHuber-NOAA <david.huber@noaa.gov>
Date: Fri, 21 Nov 2025 19:31:53 +0000
Subject: [PATCH 10/35] Correct positional arguments

---
 scripts/exgfs_atmos_postsnd.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/scripts/exgfs_atmos_postsnd.sh b/scripts/exgfs_atmos_postsnd.sh
index 559c074a0fa..5ebd42e01af 100755
--- a/scripts/exgfs_atmos_postsnd.sh
+++ b/scripts/exgfs_atmos_postsnd.sh
@@ -176,7 +176,7 @@ for (( m = 1; m <= NUM_SND_COLLECTIVES; m++ )); do
    cat "${DATA}/${m}/bufr_snd_files.list" >> "${DATA}/all_bufr_snd_files.list"
 done
 
-tar -czf "${RUN}.${cycle}.bufrsnd.tar.gz" -T "${DATA}/all_bufr_snd_files.list" -C "${COMIN_ATMOS_BUFR}"
+tar -czf "${RUN}.${cycle}.bufrsnd.tar.gz" -C "${COMIN_ATMOS_BUFR}" -T "${DATA}/all_bufr_snd_files.list"
 
 cpfs "${RUN}.${cycle}.bufrsnd.tar.gz" "${COMOUT_ATMOS_BUFR}/"
 

From ebb0e33eee42eb1ef17df8b3525ed3d4b31d75e4 Mon Sep 17 00:00:00 2001
From: DavidHuber-NOAA <david.huber@noaa.gov>
Date: Fri, 21 Nov 2025 19:50:30 +0000
Subject: [PATCH 11/35] Tar all bufr files

---
 scripts/exgfs_atmos_postsnd.sh | 37 ++++++++++++++++------------------
 1 file changed, 17 insertions(+), 20 deletions(-)

diff --git a/scripts/exgfs_atmos_postsnd.sh b/scripts/exgfs_atmos_postsnd.sh
index 5ebd42e01af..c22dc54a970 100755
--- a/scripts/exgfs_atmos_postsnd.sh
+++ b/scripts/exgfs_atmos_postsnd.sh
@@ -148,6 +148,23 @@ else
 fi
 ${runscript} "${fhr}" "${fhr_p}" "${FINT}" "${F00FLAG}" "${DATA}"
 
+############################################
+# Tar and gzip the bufr files created so far
+############################################
+find "${COMIN_ATMOS_BUFR}/bufr.*" -printf '%f\n' > "${DATA}/all_bufr_files.txt"
+
+tar -czf "${RUN}.${cycle}.bufrsnd.tar.gz" -C "${COMIN_ATMOS_BUFR}" -T "${DATA}/all_bufr_files.txt"
+
+cpfs "${RUN}.${cycle}.bufrsnd.tar.gz" "${COMOUT_ATMOS_BUFR}/"
+
+########################################
+# Send the single tar file to OSO
+########################################
+if [[ "${SENDDBN}" == "YES" ]]; then
+    "${DBNROOT}/bin/dbn_alert" MODEL GFS_BUFRSND_TAR "${job}" \
+        "${COMOUT_ATMOS_BUFR}/${RUN}.${cycle}.bufrsnd.tar.gz"
+fi
+
 ########################################
 # Create Regional Collectives of BUFR
 # data and add appropriate WMO Headers
@@ -168,26 +185,6 @@ chmod +x cmdfile
 
 ${APRUN_POSTSNDCFP} cmdfile
 
-###############################################################
-# Tar and gzip the bufr files in used to create the collectives
-###############################################################
-# The lists of files are located in ${DATA}/<m>/bufr_snd_files.list
-for (( m = 1; m <= NUM_SND_COLLECTIVES; m++ )); do
-   cat "${DATA}/${m}/bufr_snd_files.list" >> "${DATA}/all_bufr_snd_files.list"
-done
-
-tar -czf "${RUN}.${cycle}.bufrsnd.tar.gz" -C "${COMIN_ATMOS_BUFR}" -T "${DATA}/all_bufr_snd_files.list"
-
-cpfs "${RUN}.${cycle}.bufrsnd.tar.gz" "${COMOUT_ATMOS_BUFR}/"
-
-########################################
-# Send the single tar file to OSO
-########################################
-if [[ "${SENDDBN}" == "YES" ]]; then
-    "${DBNROOT}/bin/dbn_alert" MODEL GFS_BUFRSND_TAR "${job}" \
-        "${COMOUT_ATMOS_BUFR}/${RUN}.${cycle}.bufrsnd.tar.gz"
-fi
-
 
 ########################################
 # Read BUFR output and transfer into

From cdf5a8fd37bd554b413a3eb2057856b23724044c Mon Sep 17 00:00:00 2001
From: David Huber <david.huber@noaa.gov>
Date: Fri, 21 Nov 2025 22:31:17 +0000
Subject: [PATCH 12/35] Use a find command instead

---
 scripts/exgfs_atmos_postsnd.sh | 2 +-
 ush/gfs_sndp.sh                | 2 --
 2 files changed, 1 insertion(+), 3 deletions(-)

diff --git a/scripts/exgfs_atmos_postsnd.sh b/scripts/exgfs_atmos_postsnd.sh
index c22dc54a970..16b3a71e81e 100755
--- a/scripts/exgfs_atmos_postsnd.sh
+++ b/scripts/exgfs_atmos_postsnd.sh
@@ -151,7 +151,7 @@ ${runscript} "${fhr}" "${fhr_p}" "${FINT}" "${F00FLAG}" "${DATA}"
 ############################################
 # Tar and gzip the bufr files created so far
 ############################################
-find "${COMIN_ATMOS_BUFR}/bufr.*" -printf '%f\n' > "${DATA}/all_bufr_files.txt"
+find "${COMIN_ATMOS_BUFR}" -maxdepth 1 -type f -name "bufr.*" -printf '%f\n' > "${DATA}/all_bufr_files.txt"
 
 tar -czf "${RUN}.${cycle}.bufrsnd.tar.gz" -C "${COMIN_ATMOS_BUFR}" -T "${DATA}/all_bufr_files.txt"
 
diff --git a/ush/gfs_sndp.sh b/ush/gfs_sndp.sh
index 703e73845fa..98738f10c64 100755
--- a/ush/gfs_sndp.sh
+++ b/ush/gfs_sndp.sh
@@ -32,8 +32,6 @@ cd $DATA/$m
     do
        bufr_target="bufr.${stn}.${PDY}${cyc}"
        cpreq "${COMIN_ATMOS_BUFR}/${bufr_target}" "${DATA}/${m}/bufrin"
-       # Also write the filename to a list for files to tar
-       echo "${bufr_target}" >> "${DATA}/${m}/bufr_snd_files.list"
        export pgm=tocsbufr.x
        #. prep_step
        export FORT11=$DATA/${m}/bufrin

From 3303cc4f44851334a0766ce21c83bf66b4a68b41 Mon Sep 17 00:00:00 2001
From: David Huber <david.huber@noaa.gov>
Date: Tue, 25 Nov 2025 19:03:34 +0000
Subject: [PATCH 13/35] Do not load gh anymore

---
 modulefiles/gw_setup.gaeac6.lua   | 2 --
 modulefiles/gw_setup.hera.lua     | 1 -
 modulefiles/gw_setup.hercules.lua | 1 -
 modulefiles/gw_setup.orion.lua    | 1 -
 modulefiles/gw_setup.ursa.lua     | 1 -
 5 files changed, 6 deletions(-)

diff --git a/modulefiles/gw_setup.gaeac6.lua b/modulefiles/gw_setup.gaeac6.lua
index 2967cd13990..942884fd109 100644
--- a/modulefiles/gw_setup.gaeac6.lua
+++ b/modulefiles/gw_setup.gaeac6.lua
@@ -17,8 +17,6 @@ load("py-jinja2")
 load("py-pyyaml")
 load("py-numpy")
 load("git-lfs")
-prepend_path("MODULEPATH", "/ncrc/proj/epic/spack-stack/c6/spack-stack-1.9.2/envs/ue-intel-2023.2.0/install/modulefiles/gcc/12.3.0")
-load("gh")
 
 unload("cray-libsci")
 
diff --git a/modulefiles/gw_setup.hera.lua b/modulefiles/gw_setup.hera.lua
index 7e7debc259a..56f8c50e665 100644
--- a/modulefiles/gw_setup.hera.lua
+++ b/modulefiles/gw_setup.hera.lua
@@ -16,6 +16,5 @@ load(pathJoin("python", python_ver))
 load("py-jinja2")
 load("py-pyyaml")
 load("py-numpy")
-load("gh")
 
 whatis("Description: GFS run setup environment")
diff --git a/modulefiles/gw_setup.hercules.lua b/modulefiles/gw_setup.hercules.lua
index 8e9870f5b0a..fdc3a02c9d5 100644
--- a/modulefiles/gw_setup.hercules.lua
+++ b/modulefiles/gw_setup.hercules.lua
@@ -17,6 +17,5 @@ load("py-jinja2")
 load("py-pyyaml")
 load("py-numpy")
 try_load("globus-cli")
-load("gh")
 
 whatis("Description: GFS run setup environment")
diff --git a/modulefiles/gw_setup.orion.lua b/modulefiles/gw_setup.orion.lua
index c387a366f0f..1580cc57369 100644
--- a/modulefiles/gw_setup.orion.lua
+++ b/modulefiles/gw_setup.orion.lua
@@ -19,6 +19,5 @@ load("py-pyyaml")
 load("py-numpy")
 load(pathJoin("cmake", cmake_ver))
 try_load("globus-cli")
-load("gh")
 
 whatis("Description: GFS run setup environment")
diff --git a/modulefiles/gw_setup.ursa.lua b/modulefiles/gw_setup.ursa.lua
index a38c5f53e77..c7706764ff7 100644
--- a/modulefiles/gw_setup.ursa.lua
+++ b/modulefiles/gw_setup.ursa.lua
@@ -16,6 +16,5 @@ load(pathJoin("cmake", cmake_ver))
 load("py-jinja2")
 load("py-pyyaml")
 load("py-numpy")
-load("gh")
 
 whatis("Description: GFS run setup environment")

From 473c7b5860ccecd9cc50696fb6c7eed74169e1f1 Mon Sep 17 00:00:00 2001
From: David Huber <david.huber@noaa.gov>
Date: Tue, 25 Nov 2025 19:07:51 +0000
Subject: [PATCH 14/35] Enable archiving on WCOSS2 by default

---
 dev/workflow/hosts/wcoss2.yaml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/dev/workflow/hosts/wcoss2.yaml b/dev/workflow/hosts/wcoss2.yaml
index 9f82d8e9e81..2e619ce76f2 100644
--- a/dev/workflow/hosts/wcoss2.yaml
+++ b/dev/workflow/hosts/wcoss2.yaml
@@ -26,4 +26,4 @@ CHGRP_RSTPROD: 'YES'
 CHGRP_CMD: 'chgrp rstprod'
 # Features
 SUPPORTED_RESOLUTIONS: ['C1152', 'C768', 'C384', 'C192', 'C96', 'C48']
-DO_ARCHCOM: 'NO'
+DO_ARCHCOM: 'YES'

From 7cda2421bd9152674b58b0b9e0682e8d4e8c7c21 Mon Sep 17 00:00:00 2001
From: David Huber <david.huber@noaa.gov>
Date: Tue, 25 Nov 2025 19:15:12 +0000
Subject: [PATCH 15/35] Make the bufr soundings and surface files EE2-compliant

---
 ush/gfs_bfr2gpk.sh | 19 +++++++++++--------
 1 file changed, 11 insertions(+), 8 deletions(-)

diff --git a/ush/gfs_bfr2gpk.sh b/ush/gfs_bfr2gpk.sh
index 0ff2a400539..1e0ea9224aa 100755
--- a/ush/gfs_bfr2gpk.sh
+++ b/ush/gfs_bfr2gpk.sh
@@ -30,10 +30,13 @@ date
 cat "${COMOUT_ATMOS_BUFR}/bufr."*".${PDY}${cyc}" > bufr.combined
 date
 
+snd=${outfilbase}.snd
+sfc=${outfilbase}.sfc
+
 namsnd << EOF > /dev/null
 SNBUFR   = bufr.combined
-SNOUTF   = ${outfilbase}.snd
-SFOUTF   = ${outfilbase}.sfc
+SNOUTF   = ${snd}
+SFOUTF   = ${sfc}
 SNPRMF   = sngfs.prm
 SFPRMF   = sfgfs.prm
 TIMSTN   = 170/2150
@@ -46,13 +49,13 @@ date
 
 /bin/rm ./*.nts
 
-snd=${outfilbase}.snd
-sfc=${outfilbase}.sfc
-cpfs "${snd}" "${COMOUT_ATMOS_GEMPAK}/${snd}"
-cpfs "${sfc}" "${COMOUT_ATMOS_GEMPAK}/${sfc}"
+snd_out=${outfilbase}.soundings.bufr
+sfc_out=${outfilbase}.sfc.bufr
+cpfs "${snd}" "${COMOUT_ATMOS_GEMPAK}/${snd_out}"
+cpfs "${sfc}" "${COMOUT_ATMOS_GEMPAK}/${sfc_out}"
 
 if [[ ${SENDDBN} == "YES" ]]; then
-   "${DBNROOT}/bin/dbn_alert" MODEL GFS_PTYP_SFC "${job}" "${COMOUT_ATMOS_GEMPAK}/${sfc}"
-   "${DBNROOT}/bin/dbn_alert" MODEL GFS_PTYP_SND "${job}" "${COMOUT_ATMOS_GEMPAK}/${snd}"
+   "${DBNROOT}/bin/dbn_alert" MODEL GFS_PTYP_SFC "${job}" "${COMOUT_ATMOS_GEMPAK}/${sfc_out}"
+   "${DBNROOT}/bin/dbn_alert" MODEL GFS_PTYP_SND "${job}" "${COMOUT_ATMOS_GEMPAK}/${snd_out}"
 fi
 echo "done" > "${DATA}/gembufr.done"

From 41effff274d8c4950f8faea13067b6ec204a8fcf Mon Sep 17 00:00:00 2001
From: David Huber <david.huber@noaa.gov>
Date: Wed, 26 Nov 2025 14:51:44 +0000
Subject: [PATCH 16/35] Update soundings tarball name to be EE2-compliant

---
 scripts/exgfs_atmos_postsnd.sh | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

diff --git a/scripts/exgfs_atmos_postsnd.sh b/scripts/exgfs_atmos_postsnd.sh
index 16b3a71e81e..0a33ec7f9f2 100755
--- a/scripts/exgfs_atmos_postsnd.sh
+++ b/scripts/exgfs_atmos_postsnd.sh
@@ -153,16 +153,16 @@ ${runscript} "${fhr}" "${fhr_p}" "${FINT}" "${F00FLAG}" "${DATA}"
 ############################################
 find "${COMIN_ATMOS_BUFR}" -maxdepth 1 -type f -name "bufr.*" -printf '%f\n' > "${DATA}/all_bufr_files.txt"
 
-tar -czf "${RUN}.${cycle}.bufrsnd.tar.gz" -C "${COMIN_ATMOS_BUFR}" -T "${DATA}/all_bufr_files.txt"
+tar -czf "${RUN}.${cycle}.soundings.tar.gz" -C "${COMIN_ATMOS_BUFR}" -T "${DATA}/all_bufr_files.txt"
 
-cpfs "${RUN}.${cycle}.bufrsnd.tar.gz" "${COMOUT_ATMOS_BUFR}/"
+cpfs "${RUN}.${cycle}.soundings.tar.gz" "${COMOUT_ATMOS_BUFR}/"
 
 ########################################
 # Send the single tar file to OSO
 ########################################
 if [[ "${SENDDBN}" == "YES" ]]; then
     "${DBNROOT}/bin/dbn_alert" MODEL GFS_BUFRSND_TAR "${job}" \
-        "${COMOUT_ATMOS_BUFR}/${RUN}.${cycle}.bufrsnd.tar.gz"
+        "${COMOUT_ATMOS_BUFR}/${RUN}.${cycle}.soundings.tar.gz"
 fi
 
 ########################################
@@ -185,7 +185,6 @@ chmod +x cmdfile
 
 ${APRUN_POSTSNDCFP} cmdfile
 
-
 ########################################
 # Read BUFR output and transfer into
 # GEMPAK surface and sounding data files

From c8bc24ca8f2e32f3659bc78a354e301f19a24f1a Mon Sep 17 00:00:00 2001
From: David Huber <david.huber@noaa.gov>
Date: Mon, 1 Dec 2025 14:18:35 +0000
Subject: [PATCH 17/35] Correct atmos product dependencies

---
 dev/workflow/rocoto/gefs_tasks.py | 2 +-
 dev/workflow/rocoto/sfs_tasks.py  | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/dev/workflow/rocoto/gefs_tasks.py b/dev/workflow/rocoto/gefs_tasks.py
index 61438252e58..d42c0963ba6 100644
--- a/dev/workflow/rocoto/gefs_tasks.py
+++ b/dev/workflow/rocoto/gefs_tasks.py
@@ -212,7 +212,7 @@ def _atmosoceaniceprod(self, component: str):
         fhout_ice_gfs = self._configs['base']['FHOUT_ICE_GFS']
         products_dict = {'atmos': {'config': 'atmos_products',
                                    'history_path_tmpl': 'COM_ATMOS_MASTER_TMPL',
-                                   'history_file_tmpl': f'{self.run}.t@Hz.master.grb2f#fhr3_last#'},
+                                   'history_file_tmpl': f'{self.run}.t@Hz.master.f#fhr3_last#.grib2'},
                          'ocean': {'config': 'oceanice_products',
                                    'history_path_tmpl': 'COM_OCEAN_HISTORY_TMPL',
                                    'history_file_tmpl': f'{self.run}.ocean.t@Hz.{fhout_ocn_gfs}hr_avg.f#fhr3_next#.nc'},
diff --git a/dev/workflow/rocoto/sfs_tasks.py b/dev/workflow/rocoto/sfs_tasks.py
index 9113a5d9bb4..20a99af11ee 100644
--- a/dev/workflow/rocoto/sfs_tasks.py
+++ b/dev/workflow/rocoto/sfs_tasks.py
@@ -190,7 +190,7 @@ def _atmosoceaniceprod(self, component: str):
         fhout_ice_gfs = self._configs['base']['FHOUT_ICE_GFS']
         products_dict = {'atmos': {'config': 'atmos_products',
                                    'history_path_tmpl': 'COM_ATMOS_MASTER_TMPL',
-                                   'history_file_tmpl': f'{self.run}.t@Hz.master.grb2f#fhr3_last#'},
+                                   'history_file_tmpl': f'{self.run}.t@Hz.master.f#fhr3_last#.grib2'},
                          'ocean': {'config': 'oceanice_products',
                                    'history_path_tmpl': 'COM_OCEAN_HISTORY_TMPL',
                                    'history_file_tmpl': f'{self.run}.ocean.t@Hz.{fhout_ocn_gfs}hr_avg.f#fhr3_next#.nc'},

From 30395abbf5522c1407caccba7c70240b6f22398c Mon Sep 17 00:00:00 2001
From: David Huber <david.huber@noaa.gov>
Date: Mon, 1 Dec 2025 18:30:36 +0000
Subject: [PATCH 18/35] Replace tabs with spaces

---
 dev/parm/config/gfs/config.cleanup | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/dev/parm/config/gfs/config.cleanup b/dev/parm/config/gfs/config.cleanup
index 62b4aca2efd..2933a3bd010 100644
--- a/dev/parm/config/gfs/config.cleanup
+++ b/dev/parm/config/gfs/config.cleanup
@@ -28,11 +28,11 @@ fi
 # Specify the list of files to exclude from the first stage of cleanup
 # Because arrays cannot be exported, list is a single string of comma-
 # separated values. This string is split to form an array at runtime.
-export exclude_string=""
+export selective_exclude_string=""
 case ${RUN} in
-	gdas | gfs)	exclude_string+="*prepbufr*, *cnvstat.tar*, *analysis.atm.a*.nc" ;;
-	enkf*)      exclude_string+="*f006.ens*" ;;
-	*)			exclude_string="" ;;
+    gdas | gfs)	selective_exclude_string+="*prepbufr*, *cnvstat.tar*, *analysis.atm.a*.nc" ;;
+    enkf*)      selective_exclude_string+="*f006.ens*" ;;
+    *)          selective_exclude_string="" ;;
 esac
 
 echo "END: config.cleanup"

From a88f475768ee6f93ebb24cff4828db22945955a8 Mon Sep 17 00:00:00 2001
From: David Huber <david.huber@noaa.gov>
Date: Mon, 1 Dec 2025 18:31:13 +0000
Subject: [PATCH 19/35] Increase memory for gempak jobs (needed for meta task)

---
 dev/parm/config/gfs/config.resources | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/dev/parm/config/gfs/config.resources b/dev/parm/config/gfs/config.resources
index 1c8ca9d24a1..d966237b0f1 100644
--- a/dev/parm/config/gfs/config.resources
+++ b/dev/parm/config/gfs/config.resources
@@ -1380,7 +1380,7 @@ case ${step} in
     elif [[ "${RUN}" = *gfs ]]; then
       ntasks=28
       tasks_per_node=28
-      memory="4GB"
+      memory="6GB"
     fi
     ;;
 

From 159dc6823b68fa35eb3917038f84892f9a9ef3e4 Mon Sep 17 00:00:00 2001
From: David Huber <david.huber@noaa.gov>
Date: Mon, 1 Dec 2025 18:32:36 +0000
Subject: [PATCH 20/35] Make cleanup script smarter

Do selective cleanup by determining date ranges for files to keep. Only keep files based on the dates selected.
---
 scripts/exglobal_cleanup.sh | 79 +++++++++++++++++++++++--------------
 1 file changed, 50 insertions(+), 29 deletions(-)

diff --git a/scripts/exglobal_cleanup.sh b/scripts/exglobal_cleanup.sh
index 13dbc5beb32..2ecadbfb0f5 100755
--- a/scripts/exglobal_cleanup.sh
+++ b/scripts/exglobal_cleanup.sh
@@ -28,19 +28,39 @@ GEMPAK_CLEANUP_MAX=${GEMPAK_CLEANUP_MAX:-240}
 # for successful cycles (defaults from 24h to 120h).
 # Retain files needed by Fit2Obs
 first_selective_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${SELECTIVE_CLEANUP_MAX} hours")
+last_selective_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${SELECTIVE_CLEANUP_MIN} hours")
 last_rtofs_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${RTOFS_CLEANUP_MAX} hours")
-last_gempak_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${GEMPAK_CLEANUP_MAX} hours")
-exclude_string="${exclude_string:-}"
+first_gempak_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${GEMPAK_CLEANUP_MAX} hours")
+# Selective exclude list
+selective_exclude_string="${selective_exclude_string:-}"
+# Gempak exclude list
+gempak_exclude_string+=", *gfs_1p00_*"
 
-# Find the last date among all cleanup targets
+# Find the first and last date among all cleanup targets
 max_cleanup_max="${SELECTIVE_CLEANUP_MAX:-120}"
-for cleanup_max in "${RTOFS_CLEANUP_MAX}" "${GEMPAK_CLEANUP_MAX}"; do
+max_list="${RTOFS_CLEANUP_MAX}"
+if [[ "${RUN}" == "gfs" && "${DO_GEMPAK}" == "YES" ]]; then
+    max_list+=" ${GEMPAK_CLEANUP_MAX}}"
+fi
+
+for cleanup_max in ${max_list}; do
     if [[ ${cleanup_max} -gt ${max_cleanup_max} ]]; then
         max_cleanup_max=${cleanup_max}
     fi
 done
 
-last_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${max_cleanup_max} hours")
+max_cleanup_min=${max_cleanup_max}
+for cleanup_max in ${max_list}; do
+    if [[ ${cleanup_max} -lt ${max_cleanup_min} ]]; then
+        max_cleanup_min=${cleanup_max}
+    fi
+done
+
+# Start 4 cycles before the earliest exclusion target so we actually remove older files
+max_cleanup_max=$(( max_cleanup_max + 4 * assim_freq ))
+first_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${max_cleanup_max} hours")
+
+last_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${max_cleanup_min} hours")
 
 function remove_files() {
     local directory=$1
@@ -49,44 +69,45 @@ function remove_files() {
         echo "No directory ${directory} to remove files from, skiping"
         return
     fi
-    # Find all files and links in the directory and store as an arry
-    # Run find only once for efficiency
-    flist=()
-    mapfile -t flist < <(find "${directory}" -type f -or -type l) || true
-
-    # Now remove those files that match the exclude patterns
-    for exclude_pattern in "$@"; do
-        # Use a temporary array to hold files that do not match the exclude pattern
-        temp_flist=()
-        for file in "${flist[@]}"; do
-            # shellcheck disable=SC2053
-            if [[ ! $(basename "${file}") == ${exclude_pattern} ]]; then
-                temp_flist+=("${file}")
-            fi
-        done
-        flist=("${temp_flist[@]}")
-    done
-
-    # Delete all files in flist.
-    for file in "${flist[@]}"; do
-        rm -f "${file}"
+    local find_exclude_string=""
+    for exclude in "$@"; do
+        if [[ -n "${exclude}" ]]; then
+            find_exclude_string+=" -name ${exclude} -or"
+        fi
     done
+    # Chop off any trailing or
+    find_exclude_string="${find_exclude_string[*]/%-or}"
+    # Remove all regular files and symlinks that do not match
+    # shellcheck disable=SC2086
+    if [[ -n "${find_exclude_string}" ]]; then
+        # String is non-empty → use exclusion
+        find "${directory}" \( -type f -o -type l \) -not \( ${find_exclude_string} \) -ignore_readdir_race -delete
+    else
+        # String is empty → no exclusion
+        find "${directory}" \( -type f -o -type l \) -ignore_readdir_race -delete
+    fi
 
     # Remove any empty directories
     find "${directory}" -type d -empty -delete
 }
 
 # Now start removing old COM files/directories
-for (( current_date=first_selective_date; current_date <= last_date; \
+for (( current_date=first_date; current_date <= last_date; \
   current_date=$(date --utc +%Y%m%d%H -d "${current_date:0:8} ${current_date:8:2} +${assim_freq} hours") )); do
     current_PDY="${current_date:0:8}"
     current_cyc="${current_date:8:2}"
     rocotolog="${EXPDIR}/logs/${current_date}.log"
 
+    # Build the exclude list based on the 'current_date'
+    exclude_string=""
+    if [[ ${current_date} -ge ${first_selective_date} && ${current_date} -le ${last_selective_date} ]]; then
+        exclude_string+="${selective_exclude_string}"
+    fi
+
     # Extend the exclude list for gempak files if needed
-    if [[ "${RUN}" == "gfs" && ${current_date} -lt ${last_gempak_date} && "${DO_GEMPAK}" == "YES" ]]; then
+    if [[ "${RUN}" == "gfs" && ${current_date} -ge ${first_gempak_date} && "${DO_GEMPAK}" == "YES" ]]; then
         # Provide the gempak exclude pattern(s)
-        exclude_string+=", *gfs_1p00_*"
+        exclude_string+="${gempak_exclude_string}"
     fi
 
     # Check if the cycle completed successfully by looking at the rocoto log

From 73f4bcc3058acf89d33bc4797b1ce839230a28d3 Mon Sep 17 00:00:00 2001
From: David Huber <david.huber@noaa.gov>
Date: Mon, 1 Dec 2025 18:55:16 +0000
Subject: [PATCH 21/35] Reduce RTOFS retention to 24 hours

---
 dev/parm/config/gfs/config.cleanup | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/dev/parm/config/gfs/config.cleanup b/dev/parm/config/gfs/config.cleanup
index 2933a3bd010..f7be6eed83b 100644
--- a/dev/parm/config/gfs/config.cleanup
+++ b/dev/parm/config/gfs/config.cleanup
@@ -18,7 +18,7 @@ export CLEANUP_COM="YES"   # NO=retain ROTDIR.  YES default in cleanup.sh
 export SELECTIVE_CLEANUP_MAX=144
 export SELECTIVE_CLEANUP_MIN=24
 export GEMPAK_CLEANUP_MAX=240
-export RTOFS_CLEANUP_MAX=48
+export RTOFS_CLEANUP_MAX=24
 if [[ ${SELECTIVE_CLEANUP_MIN} -gt ${SELECTIVE_CLEANUP_MAX} ]]; then
     echo "FATAL ERROR: Invalid selective cleanup times: "
     echo "             SELECTIVE_CLEANUP_MIN=${SELECTIVE_CLEANUP_MIN} > SELECTIVE_CLEANUP_MAX=${SELECTIVE_CLEANUP_MAX}"

From a8d5f41c8092e30cde603fff548a799bf345674f Mon Sep 17 00:00:00 2001
From: David Huber <david.huber@noaa.gov>
Date: Tue, 2 Dec 2025 16:38:57 +0000
Subject: [PATCH 22/35] Remove extra }

---
 scripts/exglobal_cleanup.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/scripts/exglobal_cleanup.sh b/scripts/exglobal_cleanup.sh
index ddb1e1117c0..ee881962a1d 100755
--- a/scripts/exglobal_cleanup.sh
+++ b/scripts/exglobal_cleanup.sh
@@ -40,7 +40,7 @@ gempak_exclude_string+=", *gfs_1p00_*"
 max_cleanup_max="${SELECTIVE_CLEANUP_MAX:-120}"
 max_list="${RTOFS_CLEANUP_MAX}"
 if [[ "${RUN}" == "gfs" && "${DO_GEMPAK}" == "YES" ]]; then
-    max_list+=" ${GEMPAK_CLEANUP_MAX}}"
+    max_list+=" ${GEMPAK_CLEANUP_MAX}"
 fi
 
 for cleanup_max in ${max_list}; do

From cd59c0fc8ca1efc417cc7dd5284bd8feedd4f775 Mon Sep 17 00:00:00 2001
From: David Huber <david.huber@noaa.gov>
Date: Tue, 2 Dec 2025 16:58:18 +0000
Subject: [PATCH 23/35] Rename RTOFS and GEMPAK variables

---
 dev/parm/config/gfs/config.cleanup | 17 +++++++++--------
 scripts/exglobal_cleanup.sh        | 28 ++++++++++++++++------------
 2 files changed, 25 insertions(+), 20 deletions(-)

diff --git a/dev/parm/config/gfs/config.cleanup b/dev/parm/config/gfs/config.cleanup
index f7be6eed83b..61de1edb056 100644
--- a/dev/parm/config/gfs/config.cleanup
+++ b/dev/parm/config/gfs/config.cleanup
@@ -8,17 +8,18 @@ source "${EXPDIR}/config.resources" cleanup
 
 export CLEANUP_COM="YES"   # NO=retain ROTDIR.  YES default in cleanup.sh
 
-#--starting and ending hours of previous cycles to be removed from rotating directory
-# Selectively remove files between SELECTIVE_RM_MIN and SELECTIVE_RM_MIN hours old, based on exclude_string
-# Remove all RTOFS files older than RTOFS_CLEANUP_MAX hours
-# Remove all files, except GEMPAK files, older than SELECTIVE_RM_MAX hours, based on gempak_exclude_string
-# Remove all files older than the max of all SELECTIVE_RM* variables
-# Retain all files newer than SELECTIVE_RM_MIN hours
+#--starting and ending hours of previous cycles to be removed from rotating directory based on RUN
+# Selectively remove files between SELECTIVE_CLEANUP_MAX and SELECTIVE_CLEANUP_MIN hours old, based on exclude_string
+# Remove all GEMPAK files older than GEMPAK_CLEANUP hours if DO_GEMPAK is set to YES and only for RUN==gfs
+# Remove **all** files and directories older than max(SELECTIVE_CLEANUP_MAX, GEMPAK_CLEANUP, RTOFS_CLEANUP)
+
+# Keep all files newer than SELECTIVE_CLEANUP_MIN hours (except RTOFS)
+# Remove all RTOFS files older than RTOFS_CLEANUP hours
 
 export SELECTIVE_CLEANUP_MAX=144
 export SELECTIVE_CLEANUP_MIN=24
-export GEMPAK_CLEANUP_MAX=240
-export RTOFS_CLEANUP_MAX=24
+export GEMPAK_CLEANUP=240
+export RTOFS_CLEANUP=24
 if [[ ${SELECTIVE_CLEANUP_MIN} -gt ${SELECTIVE_CLEANUP_MAX} ]]; then
     echo "FATAL ERROR: Invalid selective cleanup times: "
     echo "             SELECTIVE_CLEANUP_MIN=${SELECTIVE_CLEANUP_MIN} > SELECTIVE_CLEANUP_MAX=${SELECTIVE_CLEANUP_MAX}"
diff --git a/scripts/exglobal_cleanup.sh b/scripts/exglobal_cleanup.sh
index ee881962a1d..751e0db2fb6 100755
--- a/scripts/exglobal_cleanup.sh
+++ b/scripts/exglobal_cleanup.sh
@@ -6,8 +6,8 @@ echo "Begin Cleanup ${DATAROOT}!"
 # Remove DATAoutput from the forecast model run
 # TODO: Handle this better
 DATAfcst="${DATAROOT}/${RUN}fcst.${PDY:-}${cyc}"
-if [[ -d "${DATAfcst}" ]];
-    then rm -rf "${DATAfcst}";
+if [[ -d "${DATAfcst}" ]]; then
+    rm -rf "${DATAfcst}"
 fi
 #DATAefcs="${DATAROOT}/${RUN}efcs???${PDY:-}${cyc}"
 rm -rf "${DATAROOT}/${RUN}efcs"*"${PDY:-}${cyc}"
@@ -19,8 +19,8 @@ fi
 
 SELECTIVE_CLEANUP_MIN=${SELECTIVE_CLEANUP_MIN:-24}
 SELECTIVE_CLEANUP_MAX=${SELECTIVE_CLEANUP_MAX:-120}
-RTOFS_CLEANUP_MAX=${RTOFS_CLEANUP_MAX:-48}
-GEMPAK_CLEANUP_MAX=${GEMPAK_CLEANUP_MAX:-240}
+RTOFS_CLEANUP=${RTOFS_CLEANUP:-24}
+GEMPAK_CLEANUP=${GEMPAK_CLEANUP:-240}
 ###############################################################
 # Clean up previous cycles; various depths
 
@@ -29,8 +29,8 @@ GEMPAK_CLEANUP_MAX=${GEMPAK_CLEANUP_MAX:-240}
 # Retain files needed by Fit2Obs
 first_selective_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${SELECTIVE_CLEANUP_MAX} hours")
 last_selective_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${SELECTIVE_CLEANUP_MIN} hours")
-last_rtofs_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${RTOFS_CLEANUP_MAX} hours")
-first_gempak_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${GEMPAK_CLEANUP_MAX} hours")
+last_rtofs_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${RTOFS_CLEANUP} hours")
+first_gempak_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${GEMPAK_CLEANUP} hours")
 # Selective exclude list
 selective_exclude_string="${selective_exclude_string:-}"
 # Gempak exclude list
@@ -38,9 +38,9 @@ gempak_exclude_string+=", *gfs_1p00_*"
 
 # Find the first and last date among all cleanup targets
 max_cleanup_max="${SELECTIVE_CLEANUP_MAX:-120}"
-max_list="${RTOFS_CLEANUP_MAX}"
+max_list="${RTOFS_CLEANUP}"
 if [[ "${RUN}" == "gfs" && "${DO_GEMPAK}" == "YES" ]]; then
-    max_list+=" ${GEMPAK_CLEANUP_MAX}"
+    max_list+=" ${GEMPAK_CLEANUP}"
 fi
 
 for cleanup_max in ${max_list}; do
@@ -76,7 +76,7 @@ function remove_files() {
         fi
     done
     # Chop off any trailing or
-    find_exclude_string="${find_exclude_string[*]/%-or}"
+    find_exclude_string="${find_exclude_string[*]/%-or/}"
     # Remove all regular files and symlinks that do not match
     # shellcheck disable=SC2086
     if [[ -n "${find_exclude_string}" ]]; then
@@ -92,7 +92,7 @@ function remove_files() {
 }
 
 # Now start removing old COM files/directories
-for ((current_date=first_date; current_date <= last_date; \
+for ((current_date = first_date; current_date <= last_date; \
 current_date = $(date --utc +%Y%m%d%H -d "${current_date:0:8} ${current_date:8:2} +${assim_freq} hours"))); do
     current_PDY="${current_date:0:8}"
     current_cyc="${current_date:8:2}"
@@ -124,8 +124,12 @@ current_date = $(date --utc +%Y%m%d%H -d "${current_date:0:8} ${current_date:8:2
             fi
             # Remove all rtofs directories in each RUN older than last_rtofs_date
             rtofs_dir="${ROTDIR}/rtofs.${current_PDY}"
-            if [[ -d "${rtofs_dir}" ]] && ((current_date < last_rtofs_date)); then rm -rf "${rtofs_dir}" ; fi
-            if [[ -d "${rtofs_dir}" ]] && ((current_date < last_rtofs)); then rm -rf "${rtofs_dir}"; fi
+            if [[ -d "${rtofs_dir}" && ${current_date} -lt ${last_rtofs_date} ]]; then
+                rm -rf "${rtofs_dir}"
+            fi
+            if [[ -d "${rtofs_dir}" && ${current_date} -lt ${last_rtofs} ]]; then
+                rm -rf "${rtofs_dir}"
+            fi
         fi
     fi
 done

From d2ebe7faae5eb1010be4781fb9487ee3fab0f2cf Mon Sep 17 00:00:00 2001
From: David Huber <david.huber@noaa.gov>
Date: Tue, 2 Dec 2025 17:00:07 +0000
Subject: [PATCH 24/35] Clarify RTOFS cleanup language

---
 dev/parm/config/gfs/config.cleanup | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/dev/parm/config/gfs/config.cleanup b/dev/parm/config/gfs/config.cleanup
index 61de1edb056..856a1a09413 100644
--- a/dev/parm/config/gfs/config.cleanup
+++ b/dev/parm/config/gfs/config.cleanup
@@ -8,12 +8,14 @@ source "${EXPDIR}/config.resources" cleanup
 
 export CLEANUP_COM="YES"   # NO=retain ROTDIR.  YES default in cleanup.sh
 
-#--starting and ending hours of previous cycles to be removed from rotating directory based on RUN
+#--starting and ending hours of previous cycles to be removed from rotating directory
+# RUN-based cleanup variables
+# Keep all files newer than SELECTIVE_CLEANUP_MIN hours 
 # Selectively remove files between SELECTIVE_CLEANUP_MAX and SELECTIVE_CLEANUP_MIN hours old, based on exclude_string
 # Remove all GEMPAK files older than GEMPAK_CLEANUP hours if DO_GEMPAK is set to YES and only for RUN==gfs
 # Remove **all** files and directories older than max(SELECTIVE_CLEANUP_MAX, GEMPAK_CLEANUP, RTOFS_CLEANUP)
 
-# Keep all files newer than SELECTIVE_CLEANUP_MIN hours (except RTOFS)
+# RTOFS-specific variables (runs every RUN)
 # Remove all RTOFS files older than RTOFS_CLEANUP hours
 
 export SELECTIVE_CLEANUP_MAX=144

From 93314097e122f8b607b1ec027a913e8f3322b3b5 Mon Sep 17 00:00:00 2001
From: David Huber <david.huber@noaa.gov>
Date: Tue, 2 Dec 2025 17:05:46 +0000
Subject: [PATCH 25/35] Add DO_BUFRSND dependency on gempak/bufr sounding
 outputs

---
 parm/archive/gfs_downstream.yaml.j2 | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/parm/archive/gfs_downstream.yaml.j2 b/parm/archive/gfs_downstream.yaml.j2
index 040292b7dd8..6bc663cb5b4 100644
--- a/parm/archive/gfs_downstream.yaml.j2
+++ b/parm/archive/gfs_downstream.yaml.j2
@@ -3,7 +3,7 @@ gfs_downstream:
     name: "GFS_DOWNSTREAM"
     target: "{{ ATARDIR }}/{{ cycle_YMDH }}/gfs_downstream.tar"
     required:
-        {% if DO_GEMPAK %}
+        {% if DO_GEMPAK and DO_BUFRSND %}
         - "{{ COMIN_ATMOS_GEMPAK | relpath(ROTDIR) }}/gfs_{{ cycle_YMDH }}.sfc.bufr"
         - "{{ COMIN_ATMOS_GEMPAK | relpath(ROTDIR) }}/gfs_{{ cycle_YMDH }}.soundings.bufr"
         {% endif %}

From 19e44a1fadf2bc070e424739a47bddc1940c9e6c Mon Sep 17 00:00:00 2001
From: David Huber <david.huber@noaa.gov>
Date: Tue, 2 Dec 2025 17:10:07 +0000
Subject: [PATCH 26/35] Linter cleanup

---
 scripts/exglobal_cleanup.sh | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/scripts/exglobal_cleanup.sh b/scripts/exglobal_cleanup.sh
index 751e0db2fb6..279b7a514dc 100755
--- a/scripts/exglobal_cleanup.sh
+++ b/scripts/exglobal_cleanup.sh
@@ -57,7 +57,7 @@ for cleanup_max in ${max_list}; do
 done
 
 # Start 4 cycles before the earliest exclusion target so we actually remove older files
-max_cleanup_max=$(( max_cleanup_max + 4 * assim_freq ))
+max_cleanup_max=$((max_cleanup_max + 4 * assim_freq))
 first_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${max_cleanup_max} hours")
 
 last_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${max_cleanup_min} hours")
@@ -92,7 +92,7 @@ function remove_files() {
 }
 
 # Now start removing old COM files/directories
-for ((current_date = first_date; current_date <= last_date; \
+for ((current_date = first_date; current_date <= last_date;  \
 current_date = $(date --utc +%Y%m%d%H -d "${current_date:0:8} ${current_date:8:2} +${assim_freq} hours"))); do
     current_PDY="${current_date:0:8}"
     current_cyc="${current_date:8:2}"

From 844d5280d98267bf7a252c268655675222e4028e Mon Sep 17 00:00:00 2001
From: David Huber <david.huber@noaa.gov>
Date: Tue, 2 Dec 2025 18:53:22 +0000
Subject: [PATCH 27/35] Rename variables to more meaningful names

---
 scripts/exglobal_cleanup.sh | 15 +++++++--------
 1 file changed, 7 insertions(+), 8 deletions(-)

diff --git a/scripts/exglobal_cleanup.sh b/scripts/exglobal_cleanup.sh
index 279b7a514dc..9c6d5d09331 100755
--- a/scripts/exglobal_cleanup.sh
+++ b/scripts/exglobal_cleanup.sh
@@ -29,16 +29,18 @@ GEMPAK_CLEANUP=${GEMPAK_CLEANUP:-240}
 # Retain files needed by Fit2Obs
 first_selective_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${SELECTIVE_CLEANUP_MAX} hours")
 last_selective_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${SELECTIVE_CLEANUP_MIN} hours")
-last_rtofs_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${RTOFS_CLEANUP} hours")
-first_gempak_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${GEMPAK_CLEANUP} hours")
+gempak_cutoff_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${GEMPAK_CLEANUP} hours")
 # Selective exclude list
 selective_exclude_string="${selective_exclude_string:-}"
 # Gempak exclude list
 gempak_exclude_string+=", *gfs_1p00_*"
 
+# RTOFS is cleaned separately (for all RUNs)
+rtofs_cutoff_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${RTOFS_CLEANUP} hours")
+
 # Find the first and last date among all cleanup targets
 max_cleanup_max="${SELECTIVE_CLEANUP_MAX:-120}"
-max_list="${RTOFS_CLEANUP}"
+max_list="${SELECTIVE_CLEANUP_MAX}"
 if [[ "${RUN}" == "gfs" && "${DO_GEMPAK}" == "YES" ]]; then
     max_list+=" ${GEMPAK_CLEANUP}"
 fi
@@ -105,7 +107,7 @@ current_date = $(date --utc +%Y%m%d%H -d "${current_date:0:8} ${current_date:8:2
     fi
 
     # Extend the exclude list for gempak files if needed
-    if [[ "${RUN}" == "gfs" && ${current_date} -ge ${first_gempak_date} && "${DO_GEMPAK}" == "YES" ]]; then
+    if [[ "${RUN}" == "gfs" && ${current_date} -ge ${gempak_cutoff_date} && "${DO_GEMPAK}" == "YES" ]]; then
         # Provide the gempak exclude pattern(s)
         exclude_string+="${gempak_exclude_string}"
     fi
@@ -124,10 +126,7 @@ current_date = $(date --utc +%Y%m%d%H -d "${current_date:0:8} ${current_date:8:2
             fi
             # Remove all rtofs directories in each RUN older than last_rtofs_date
             rtofs_dir="${ROTDIR}/rtofs.${current_PDY}"
-            if [[ -d "${rtofs_dir}" && ${current_date} -lt ${last_rtofs_date} ]]; then
-                rm -rf "${rtofs_dir}"
-            fi
-            if [[ -d "${rtofs_dir}" && ${current_date} -lt ${last_rtofs} ]]; then
+            if [[ -d "${rtofs_dir}" && ${current_date} -lt ${rtofs_cutoff_date} ]]; then
                 rm -rf "${rtofs_dir}"
             fi
         fi

From 28659b91a5f06e937e23a3868622237e4530c903 Mon Sep 17 00:00:00 2001
From: David Huber <david.huber@noaa.gov>
Date: Tue, 2 Dec 2025 19:51:08 +0000
Subject: [PATCH 28/35] Only delete empty RUN.PDY directories

---
 scripts/exglobal_cleanup.sh | 17 ++++++-----------
 1 file changed, 6 insertions(+), 11 deletions(-)

diff --git a/scripts/exglobal_cleanup.sh b/scripts/exglobal_cleanup.sh
index 9c6d5d09331..1e5d51f8c78 100755
--- a/scripts/exglobal_cleanup.sh
+++ b/scripts/exglobal_cleanup.sh
@@ -33,7 +33,7 @@ gempak_cutoff_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${GEMPAK_CLEANUP} h
 # Selective exclude list
 selective_exclude_string="${selective_exclude_string:-}"
 # Gempak exclude list
-gempak_exclude_string+=", *gfs_1p00_*"
+gempak_exclude_string+=", gfs_1p00_*"
 
 # RTOFS is cleaned separately (for all RUNs)
 rtofs_cutoff_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${RTOFS_CLEANUP} hours")
@@ -51,19 +51,14 @@ for cleanup_max in ${max_list}; do
     fi
 done
 
-max_cleanup_min=${max_cleanup_max}
-for cleanup_max in ${max_list}; do
-    if [[ ${cleanup_max} -lt ${max_cleanup_min} ]]; then
-        max_cleanup_min=${cleanup_max}
-    fi
-done
-
 # Start 4 cycles before the earliest exclusion target so we actually remove older files
 max_cleanup_max=$((max_cleanup_max + 4 * assim_freq))
 first_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${max_cleanup_max} hours")
 
-last_date=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${max_cleanup_min} hours")
+# Cleanup starts on SELECTIVE_CLEANUP_MIN
+last_date=${last_selective_date}
 
+# Declare remove_files(), which will be called in a loop to selectively remove files
 function remove_files() {
     local directory=$1
     shift
@@ -158,7 +153,7 @@ if [[ "${RUN}" == "gfs" ]]; then
     done
 fi
 
-# Remove $RUN.$rPDY for the older of GDATE or RDATE
+# Remove $RUN.$rPDY for the older of GDATE or RDATE if it is empty
 GDATE=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${max_cleanup_max} hours")
 RDATE=$(date --utc +%Y%m%d%H -d "${PDY} ${cyc} -${FHMAX_GFS} hours")
 if ((GDATE < RDATE)); then
@@ -167,7 +162,7 @@ fi
 
 deletion_target="${ROTDIR}/${RUN}.${RDATE:0:8}"
 if [[ -d "${deletion_target}" ]]; then
-    rm -rf "${deletion_target}"
+    find "${deletion_target}" -maxdepth 0 -type d -empty -delete
 fi
 
 # sync and wait to avoid filesystem synchronization issues

From 38793d2610d500124fb523a3fd1c6a2df99e7492 Mon Sep 17 00:00:00 2001
From: David Huber <david.huber@noaa.gov>
Date: Tue, 2 Dec 2025 19:51:39 +0000
Subject: [PATCH 29/35] Add check for invalid GEMPAK retention

---
 dev/parm/config/gfs/config.cleanup | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/dev/parm/config/gfs/config.cleanup b/dev/parm/config/gfs/config.cleanup
index 856a1a09413..53c4a0192f4 100644
--- a/dev/parm/config/gfs/config.cleanup
+++ b/dev/parm/config/gfs/config.cleanup
@@ -28,6 +28,13 @@ if [[ ${SELECTIVE_CLEANUP_MIN} -gt ${SELECTIVE_CLEANUP_MAX} ]]; then
     exit 1
 fi
 
+# GEMPAK_CLEANUP must not be less than SELECTIVE_CLEANUP_MIN, which acts as the starting point for cleanup
+if [[ "${DO_GEMPAK}" == "YES" && ${GEMPAK_CLEANUP} -lt ${GEMPAK_CLEANUP} ]]; then
+    echo "FATAL ERROR: Invalid GEMPAK cleanup time: "
+    echo "             GEMPAK_CLEANUP=${GEMPAK_CLEANUP} < SELECTIVE_CLEANUP_MIN=${SELECTIVE_CLEANUP_MIN}"
+    exit 2
+fi
+
 # Specify the list of files to exclude from the first stage of cleanup
 # Because arrays cannot be exported, list is a single string of comma-
 # separated values. This string is split to form an array at runtime.

From bb7a31a655278a22d094862c285f40fb1871990e Mon Sep 17 00:00:00 2001
From: David Huber <david.huber@noaa.gov>
Date: Tue, 9 Dec 2025 18:29:46 +0000
Subject: [PATCH 30/35] Re-disable HPSS archiving on WCOSS2 by default

---
 dev/workflow/hosts/wcoss2.yaml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/dev/workflow/hosts/wcoss2.yaml b/dev/workflow/hosts/wcoss2.yaml
index 2e619ce76f2..9f82d8e9e81 100644
--- a/dev/workflow/hosts/wcoss2.yaml
+++ b/dev/workflow/hosts/wcoss2.yaml
@@ -26,4 +26,4 @@ CHGRP_RSTPROD: 'YES'
 CHGRP_CMD: 'chgrp rstprod'
 # Features
 SUPPORTED_RESOLUTIONS: ['C1152', 'C768', 'C384', 'C192', 'C96', 'C48']
-DO_ARCHCOM: 'YES'
+DO_ARCHCOM: 'NO'

From 4a8bb34ac7e8e6ac9c19911f2c0fa55e23736d2d Mon Sep 17 00:00:00 2001
From: David Huber <david.huber@noaa.gov>
Date: Tue, 9 Dec 2025 18:30:33 +0000
Subject: [PATCH 31/35] Disable C48_S2SW on WCOSS2

This test is covered by C48_S2SW_extended
---
 dev/ci/cases/pr/C48_S2SW.yaml | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/dev/ci/cases/pr/C48_S2SW.yaml b/dev/ci/cases/pr/C48_S2SW.yaml
index 59148444792..cd7e5875903 100644
--- a/dev/ci/cases/pr/C48_S2SW.yaml
+++ b/dev/ci/cases/pr/C48_S2SW.yaml
@@ -11,6 +11,9 @@ experiment:
   edate: 2021032312
   yaml: {{ HOMEgfs }}/dev/ci/cases/yamls/gfs_defaults_ci.yaml
 
+skip_ci_on_hosts:
+  - wcoss2
+
 workflow:
   engine: rocoto
   rocoto:

From 8382dea4ca18982bf6bb488e153ef02fc347228e Mon Sep 17 00:00:00 2001
From: David Huber <david.huber@noaa.gov>
Date: Tue, 9 Dec 2025 18:31:03 +0000
Subject: [PATCH 32/35] Turn on HPSS archiving for extended tests

---
 dev/ci/cases/yamls/gfs_extended_ci.yaml | 1 +
 1 file changed, 1 insertion(+)

diff --git a/dev/ci/cases/yamls/gfs_extended_ci.yaml b/dev/ci/cases/yamls/gfs_extended_ci.yaml
index bbf8155ac3f..b45ac7d935a 100644
--- a/dev/ci/cases/yamls/gfs_extended_ci.yaml
+++ b/dev/ci/cases/yamls/gfs_extended_ci.yaml
@@ -13,3 +13,4 @@ base:
   FHMAX_GFS: 384
   FHMAX_HF_GFS: 120
   DO_TEST_MODE: "NO"
+  DO_ARCHCOM: "YES"

From fb0f7448fe45b65b400e87b242d85e97bb6c166f Mon Sep 17 00:00:00 2001
From: David Huber <david.huber@noaa.gov>
Date: Tue, 9 Dec 2025 19:00:28 +0000
Subject: [PATCH 33/35] Revert the use of 'bufr_target'

---
 ush/gfs_sndp.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ush/gfs_sndp.sh b/ush/gfs_sndp.sh
index 98738f10c64..47de659c37a 100755
--- a/ush/gfs_sndp.sh
+++ b/ush/gfs_sndp.sh
@@ -31,7 +31,7 @@ cd $DATA/$m
     for stn in $(cat $file_list)
     do
        bufr_target="bufr.${stn}.${PDY}${cyc}"
-       cpreq "${COMIN_ATMOS_BUFR}/${bufr_target}" "${DATA}/${m}/bufrin"
+       cpreq "${COMIN_ATMOS_BUFR}/bufr.${stn}.${PDY}${cyc}" "${DATA}/${m}/bufrin"
        export pgm=tocsbufr.x
        #. prep_step
        export FORT11=$DATA/${m}/bufrin

From 246487ca68f9d2496963fc01d0a38f8b10be9edf Mon Sep 17 00:00:00 2001
From: David Huber <david.huber@noaa.gov>
Date: Tue, 9 Dec 2025 19:01:29 +0000
Subject: [PATCH 34/35] Remove now-unused variable

---
 ush/gfs_sndp.sh | 1 -
 1 file changed, 1 deletion(-)

diff --git a/ush/gfs_sndp.sh b/ush/gfs_sndp.sh
index 47de659c37a..af09e175dd7 100755
--- a/ush/gfs_sndp.sh
+++ b/ush/gfs_sndp.sh
@@ -30,7 +30,6 @@ cd $DATA/$m
 
     for stn in $(cat $file_list)
     do
-       bufr_target="bufr.${stn}.${PDY}${cyc}"
        cpreq "${COMIN_ATMOS_BUFR}/bufr.${stn}.${PDY}${cyc}" "${DATA}/${m}/bufrin"
        export pgm=tocsbufr.x
        #. prep_step

From 865a53d2e134f74232f82e6d0e98f9637714c8d9 Mon Sep 17 00:00:00 2001
From: David Huber <david.huber@noaa.gov>
Date: Tue, 9 Dec 2025 19:21:37 +0000
Subject: [PATCH 35/35] Correct GEMPAK_CLEANUP check

---
 dev/parm/config/gfs/config.cleanup | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/dev/parm/config/gfs/config.cleanup b/dev/parm/config/gfs/config.cleanup
index 53c4a0192f4..a4af66b493d 100644
--- a/dev/parm/config/gfs/config.cleanup
+++ b/dev/parm/config/gfs/config.cleanup
@@ -29,7 +29,7 @@ if [[ ${SELECTIVE_CLEANUP_MIN} -gt ${SELECTIVE_CLEANUP_MAX} ]]; then
 fi
 
 # GEMPAK_CLEANUP must not be less than SELECTIVE_CLEANUP_MIN, which acts as the starting point for cleanup
-if [[ "${DO_GEMPAK}" == "YES" && ${GEMPAK_CLEANUP} -lt ${GEMPAK_CLEANUP} ]]; then
+if [[ "${DO_GEMPAK}" == "YES" && ${GEMPAK_CLEANUP} -lt ${SELECTIVE_CLEANUP_MIN} ]]; then
     echo "FATAL ERROR: Invalid GEMPAK cleanup time: "
     echo "             GEMPAK_CLEANUP=${GEMPAK_CLEANUP} < SELECTIVE_CLEANUP_MIN=${SELECTIVE_CLEANUP_MIN}"
     exit 2