From 7105b984de14bff658c4fa4aead79375ae3b13d8 Mon Sep 17 00:00:00 2001
From: Copilot <198982749+Copilot@users.noreply.github.com>
Date: Tue, 9 Dec 2025 15:00:01 -0600
Subject: [PATCH 1/2] ospool-image-puller: Use argparse; add the --keep-days
 argument for choosing how many days of images to keep

(Copilot, GPT-5 mini)
---
 opensciencegrid/ospool-image-puller/update.py | 40 +++++++++++++------
 1 file changed, 27 insertions(+), 13 deletions(-)

diff --git a/opensciencegrid/ospool-image-puller/update.py b/opensciencegrid/ospool-image-puller/update.py
index 813b4e7a..554dc455 100755
--- a/opensciencegrid/ospool-image-puller/update.py
+++ b/opensciencegrid/ospool-image-puller/update.py
@@ -1,22 +1,37 @@
 #!/usr/bin/env python3
 
 import os
-import sys
 import time
 import yaml
 import datetime
 import subprocess
+import argparse
 
 start_dir = os.getcwd()
 
 def main():
-    if len(sys.argv) != 2:
-        print("Usage: update.py <target_dir>")
-        sys.exit(1)
-
-    target_dir = sys.argv[1]
-
-    with open("images.yaml") as file:
+    parser = argparse.ArgumentParser(
+        description='Pull and update container images from Docker to Apptainer/Singularity format'
+    )
+    parser.add_argument(
+        'target_dir',
+        help='Target directory where images will be stored'
+    )
+    parser.add_argument(
+        '--config',
+        default='images.yaml',
+        help='Path to the configuration YAML file (default: images.yaml)'
+    )
+    parser.add_argument(
+        '--keep-days',
+        type=int,
+        default=10,
+        help='Number of days to keep old images (default: 10)'
+    )
+
+    args = parser.parse_args()
+
+    with open(args.config) as file:
         conf = yaml.safe_load(file)
 
     now = datetime.datetime.now(datetime.timezone.utc)
@@ -36,13 +51,13 @@ def main():
             sing_arch = "arm64" if arch == "aarch64" else "amd64"
 
             os.chdir(start_dir)
-            os.makedirs(os.path.join(target_dir, arch, img["name"]), exist_ok=True)
-            os.chdir(os.path.join(target_dir, arch, img["name"]))
+            os.makedirs(os.path.join(args.target_dir, arch, img["name"]), exist_ok=True)
+            os.chdir(os.path.join(args.target_dir, arch, img["name"]))
 
             print(f"Working in {os.getcwd()}")
 
             # log the build to a file in the same structure under logs/
-            log_dir = os.path.join("../../..", target_dir, "logs", arch)
+            log_dir = os.path.join("../../..", args.target_dir, "logs", arch)
             os.makedirs(log_dir, exist_ok=True)
             log_file = os.path.join(log_dir, f"{img['name']}.txt")
 
@@ -65,7 +80,7 @@ def main():
             subprocess.run("ls *.sif | sort | tail -n 1 > latest.txt", shell=True)
 
             # cleanup old images, keep only latest N
-            subprocess.run("find . -maxdepth 1 -name \\*.sif -mtime +10 -exec rm -f {} \\;", shell=True)
+            subprocess.run(f"find . -maxdepth 1 -name \\*.sif -mtime +{args.keep_days} -exec rm -f {{}} \\;", shell=True)
 
     now2 = datetime.datetime.now(datetime.timezone.utc)
     now2_human = now2.strftime("%Y-%m-%d %H:%M:%S %Z")
@@ -76,4 +91,3 @@ def main():
 
 if __name__ == "__main__":
     main()
-

From e759ea6000cda63e44916662da2decbe87d4de16 Mon Sep 17 00:00:00 2001
From: Matyas Selmeci <mselmeci@wisc.edu>
Date: Tue, 9 Dec 2025 15:07:24 -0600
Subject: [PATCH 2/2] ospool-image-puller: Add KEEP_DAYS parameter to
 Dockerfile and cron job

---
 opensciencegrid/ospool-image-puller/Dockerfile | 2 +-
 opensciencegrid/ospool-image-puller/startup.sh | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/opensciencegrid/ospool-image-puller/Dockerfile b/opensciencegrid/ospool-image-puller/Dockerfile
index fe8a41b5..893a2e48 100644
--- a/opensciencegrid/ospool-image-puller/Dockerfile
+++ b/opensciencegrid/ospool-image-puller/Dockerfile
@@ -18,8 +18,8 @@ COPY --chmod=0755 startup.sh /bin/startup.sh
 ENV CRON_EXPR="30 5 * * *"
 ENV TIMEOUT="23h"
 ENV IMAGES_UID="1000"
+ENV KEEP_DAYS=10
 
 VOLUME ["/ospool/images"]
 
 CMD ["/bin/startup.sh"]
-
diff --git a/opensciencegrid/ospool-image-puller/startup.sh b/opensciencegrid/ospool-image-puller/startup.sh
index 9828b35f..9f3e891d 100755
--- a/opensciencegrid/ospool-image-puller/startup.sh
+++ b/opensciencegrid/ospool-image-puller/startup.sh
@@ -16,7 +16,7 @@ install -o images -d "${TARGET_DIR}"
 # the container's stdout/err, but then drop privileges to run the
 # actual script.
 
-echo "${CRON_EXPR?} root cd /ospool/images-scripts && /usr/sbin/runuser -u images -- timeout -k 60s ${TIMEOUT?} ./update.py ${TARGET_DIR}" '> /proc/1/fd/1 2>&1' > /etc/cron.d/update.cron
+echo "${CRON_EXPR?} root cd /ospool/images-scripts && /usr/sbin/runuser -u images -- timeout -k 60s ${TIMEOUT?} ./update.py --keep-days=${KEEP_DAYS:-10} ${TARGET_DIR}" '> /proc/1/fd/1 2>&1' > /etc/cron.d/update.cron
 
 # Start cron in non-daemon (foreground) mode
 echo >&2 "Starting cron"