Flush transitive aliases during deployment [RHELDST-28321]

We saw some repo instability after a recent GA. This was due to only a
portion of affected URLs being flushed. The current deployment cache
flush only gets URLs based on directly updated aliases. This didn't take
into consideration URLs that require several aliases to be applied.

exodus_gw/models/path.py

@@ -5,7 +5,7 @@
 
 from .base import Base
 
-
+# pylint: disable=unsubscriptable-object
 class PublishedPath(Base):
     """Represents a path updated on the CDN at some point.
 

exodus_gw/models/publish.py

@@ -21,7 +21,7 @@
 
 from .base import Base
 
-
+# pylint: disable=unsubscriptable-object
 class Publish(Base):
     __tablename__ = "publishes"
 

exodus_gw/worker/deploy.py

@@ -8,6 +8,7 @@
 
 from exodus_gw import models, schemas
 from exodus_gw.aws.dynamodb import DynamoDB
+from exodus_gw.aws.util import uris_with_aliases
 from exodus_gw.database import db_engine
 from exodus_gw.settings import Settings
 
@@ -144,30 +145,40 @@ def deploy_config(
     # URLs depending on what changed in the config.
     flush_paths: set[str] = set()
 
+    updated_prefixes = set()
     for src, updated_dest, _ in ddb.aliases_for_flush:
         if original_aliases.get(src) != updated_dest:
-            for published_path in db.query(models.PublishedPath).filter(
-                models.PublishedPath.env == env,
-                models.PublishedPath.web_uri.like(f"{src}/%"),
+            updated_prefixes.add(src)
+
+    aliases_to_expand = [alias for alias in ddb.aliases_for_flush
+                         if alias[0] in original_aliases.keys()]
+    updated_prefixes.update(uris_with_aliases(updated_prefixes, aliases_to_expand))
+
+    for src in updated_prefixes:
+        for published_path in db.query(models.PublishedPath).filter(
+            models.PublishedPath.env == env,
+            models.PublishedPath.web_uri.like(f"{src}/%"),
+        ):
+            # If any original exclusion matches the uri, the uri wouldn't
+            # have been treated as an alias, thus cache flushing would be
+            # unnecessary.
+            if any(
+                [
+                    re.search(exclusion, published_path.web_uri)
+                    for exclusion in original_exclusions.get(src, [])
+                ]
             ):
-                # If any original exclusion matches the uri, the uri wouldn't
-                # have been treated as an alias, thus cache flushing would be
-                # unnecessary.
-                if any(
-                    [
-                        re.search(exclusion, published_path.web_uri)
-                        for exclusion in original_exclusions.get(src, [])
-                    ]
-                ):
-                    continue
-                LOG.info(
-                    "Updated alias %s will flush cache for %s",
-                    src,
-                    published_path.web_uri,
-                    extra={"event": "deploy"},
-                )
-                flush_paths.add(published_path.web_uri)
-
+                continue
+            LOG.info(
+                "Updated alias %s will flush cache for %s",
+                src,
+                published_path.web_uri,
+                extra={"event": "deploy"},
+            )
+            flush_paths.add(published_path.web_uri)
+
+    # Need to expand the matched aliases to other aliases that'd be populated
+    # during publish
     # Include all the listing paths for flush when enabled in settings
     flush_paths = (
         flush_paths.union(_listing_paths_for_flush(config))

tests/worker/test_deploy.py

@@ -158,6 +158,21 @@ def test_deploy_config_with_flush(
         )
     )
 
+    # These paths should be flushed after expanding the updated aliases.
+    db.add(
+        PublishedPath(
+            env="test",
+            web_uri="/content/testproduct/rhui/1/file1",
+            updated=datetime.now(tz=timezone.utc),
+        )
+    )
+    db.add(
+        PublishedPath(
+            env="test",
+            web_uri="/content/testproduct/rhui/1/file2",
+            updated=datetime.now(tz=timezone.utc),
+        )
+    )
     db.commit()
 
     # We're updating the alias in the config.
@@ -223,6 +238,8 @@ def test_deploy_config_with_flush(
         "/content/testproduct/1/file1",
         "/content/testproduct/1/file2",
         "/content/testproduct/1/newExclusion/file5",
+        "/content/testproduct/rhui/1/file1",
+        "/content/testproduct/rhui/1/file2",
     ]
 
     # And actor call should have been delayed by this long.