From 0f9feaf51c06a4f86ab695269f9506cb86f95556 Mon Sep 17 00:00:00 2001
From: Jeremy Cline <jeremycline@linux.microsoft.com>
Date: Feb 03 2025 16:33:26 +0000
Subject: aws: Convert WaiterError to a Nack and bump wait times


Occasionally an AWS region is slow and fails an import/copy before the
timeout (30 minutes). This exception wasn't handled in the outer
exception handling block which led to the fedora_messaging consumer
catching and logging the error.

This bumps the timeout to 1 hour and adds an except block to log and
Nack a message when the timeout is hit.

Fixes #36

Signed-off-by: Jeremy Cline <jeremycline@linux.microsoft.com>

---

diff --git a/fedora-image-uploader/fedora_image_uploader/aws.py b/fedora-image-uploader/fedora_image_uploader/aws.py
index cb85247..4fd9a31 100644
--- a/fedora-image-uploader/fedora_image_uploader/aws.py
+++ b/fedora-image-uploader/fedora_image_uploader/aws.py
@@ -88,6 +88,10 @@ class Aws:
                 else:
                     raise
             regions_to_amis = self.aws_copy_image_to_regions(image, ffrel, image_id, ami_name)
+        except boto_exceptions.WaiterError as e:
+            # Typically this is because the region is slow; we'll retry.
+            _log.warning("Failed while waiting for AWS image registration: %s", str(e))
+            raise fm_exceptions.Nack()
         except boto_exceptions.ClientError as e:
             _log.warning("Failed to import image: %s", str(e))
             raise fm_exceptions.Nack()
@@ -152,7 +156,7 @@ class Aws:
         snapshot_import_waiter = self.ec2_client.get_waiter("snapshot_imported")
         snapshot_import_waiter.wait(
             ImportTaskIds=[snapshot_import_task["ImportTaskId"]],
-            WaiterConfig={"Delay": 30, "MaxAttempts": 60},
+            WaiterConfig={"Delay": 30, "MaxAttempts": 120},
         )
         snapshot_imports = self.ec2_client.describe_import_snapshot_tasks(
             ImportTaskIds=[snapshot_import_task["ImportTaskId"]],
@@ -264,7 +268,10 @@ class Aws:
         # Wait for EC2 to mark the image as available before returning since copying
         # to other regions will fail if the image isn't available.
         waiter = self.ec2_client.get_waiter("image_available")
-        waiter.wait(ImageIds=[aws_image_id])
+        waiter.wait(
+            ImageIds=[aws_image_id],
+            WaiterConfig={"Delay": 30, "MaxAttempts": 120},
+        )
         return aws_image_id
 
     def aws_copy_image_to_regions(
@@ -323,7 +330,10 @@ class Aws:
         for region, image_id in image_ids.items():
             region_client = boto3.client("ec2", region_name=region, config=self.boto_config)
             waiter = region_client.get_waiter("image_available")
-            waiter.wait(ImageIds=[image_id])
+            waiter.wait(
+                ImageIds=[image_id],
+                WaiterConfig={"Delay": 30, "MaxAttempts": 120},
+            )
             _log.info("Image finished copying to %s as %s, marking as public", region, image_id)
             try:
                 region_client.modify_image_attribute(