Add flag to retain recently used images

.gitignore

@@ -7,3 +7,4 @@ build/
 __pycache__
 
 *.egg-info/
+.venv

docker_custodian/docker_gc.py

@@ -7,6 +7,7 @@
 import fnmatch
 import logging
 import sys
+import time
 
 import dateutil.parser
 import docker
@@ -137,7 +138,13 @@ def get_dangling_volumes(client):
     return volumes
 
 
-def cleanup_images(client, max_image_age, dry_run, exclude_set):
+def cleanup_images(
+    client,
+    max_image_age,
+    recent_image_age,
+    dry_run,
+    exclude_set
+):
     # re-fetch container list so that we don't include removed containers
 
     containers = get_all_containers(client)
@@ -151,10 +158,35 @@ def cleanup_images(client, max_image_age, dry_run, exclude_set):
         images = filter_images_in_use_by_id(images, image_ids_in_use)
     images = filter_excluded_images(images, exclude_set)
 
+    if recent_image_age is not None:
+        images = without_recently_used_images(client, images, recent_image_age)
+
     for image_summary in reversed(list(images)):
         remove_image(client, image_summary, max_image_age, dry_run)
 
 
+def without_recently_used_images(client, images, recent_image_age):
+    exclude = set()
+    for event in client.events(since=recent_image_age,
+                               until=int(time.time()),
+                               decode=True):
+        status = event.get('status', '')
+        if status != 'create' and not status.startswith('exec_start'):
+            continue
+        image_ref = event.get('from')
+        if image_ref is None:
+            continue
+        exclude.add(image_ref)
+    return [
+        image for image in images
+        if exclude.intersection(
+                [image.get('Id')] +
+                image.get('RepoTags', []) +
+                image.get('RepoDigests', [])
+        )
+    ]
+
+
 def filter_excluded_images(images, exclude_set):
     def include_image(image_summary):
         image_tags = image_summary.get('RepoTags')
@@ -314,7 +346,13 @@ def main():
         exclude_set = build_exclude_set(
             args.exclude_image,
             args.exclude_image_file)
-        cleanup_images(client, args.max_image_age, args.dry_run, exclude_set)
+        cleanup_images(
+            client,
+            args.max_image_age,
+            args.recent_image_age,
+            args.dry_run,
+            exclude_set
+        )
 
     if args.dangling_volumes:
         cleanup_volumes(client, args.dry_run)
@@ -334,6 +372,11 @@ def get_args(args=None):
         help="Maxium age for an image. Images older than this age will be "
              "removed. Age can be specified in any pytimeparse supported "
              "format.")
+    parser.add_argument(
+        '--recent-image-age',
+        type=timedelta_type, default=None,
+        help="Images used within specified time interval will not be removed. "
+             "Age can be specified in any pytimeparse supported format.")
     parser.add_argument(
         '--dangling-volumes',
         action="store_true",

tests/docker_gc_test.py

@@ -124,12 +124,30 @@ def test_cleanup_images(mock_client, now):
     ]
     mock_client.inspect_image.side_effect = iter(mock_images)
 
-    docker_gc.cleanup_images(mock_client, max_image_age, False, set())
+    docker_gc.cleanup_images(mock_client, max_image_age, None, False, set())
     assert mock_client.remove_image.mock_calls == [
         mock.call(image=image['Id']) for image in reversed(images)
     ]
 
 
+def test_without_recently_used_images(mock_client, now):
+    images = [
+        dict(Id='gone'),
+        dict(Id='a'),
+        dict(RepoTags=['b']),
+        dict(RepoDigests=['c'])
+    ]
+    mock_events = [
+        {'status': 'create', 'from': 'a'},
+        {'status': 'exec_start', 'from': 'b'},
+        {'status': 'create', 'from': 'c'}
+    ]
+    mock_client.events.return_value = mock_events
+    filtered_images = docker_gc.without_recently_used_images(
+        mock_client, images, 0)
+    assert filtered_images == images[1:]
+
+
 def test_cleanup_volumes(mock_client):
     mock_client.volumes.return_value = volumes = {
         'Volumes': [
@@ -238,7 +256,7 @@ def test_filter_images_in_use_by_id(mock_client, now):
         'Id': image,
         'Created': '2014-01-01T01:01:01Z'
     }
-    docker_gc.cleanup_images(mock_client, now, False, set())
+    docker_gc.cleanup_images(mock_client, now, None, False, set())
     assert mock_client.remove_image.mock_calls == [
         mock.call(image=id_) for id_ in ['6', '5', '4', '3']
     ]