rgw: check for duplicate rule priorities in replication configuration

When multiple replication rules target the same bucket, the rule
with the highest priority should be applied. While this logic is
already in place, the configuration does not enforce priority
settings, which could lead to confusion. This behavior aligns with
AWS's implementation for multiple destination buckets as outlined
https://aws.amazon.com/blogs/aws/new-amazon-s3-replication-adds-support-for-multiple-destination-buckets/

Fixes: https://tracker.ceph.com/issues/69946
Signed-off-by: Seena Fallah <seenafallah@gmail.com>

diff --git a/src/rgw/rgw_rest_s3.cc b/src/rgw/rgw_rest_s3.cc
index d2ce7ca6afe210f386433e8abdd071294016c7db..e7ef09b5aa1f4fef1166b58b0c93ce31a6c0dd37 100644 (file)
--- a/src/rgw/rgw_rest_s3.cc
+++ b/src/rgw/rgw_rest_s3.cc
@@ -1419,7 +1419,13 @@ struct ReplicationConfiguration {
     disabled_group.id = disabled_group_id;
     disabled_group.status = rgw_sync_policy_group::Status::ALLOWED; /* not enabled, not forbidden */
 
+    std::unordered_set<int32_t> priorities; // used to check for duplicates
     for (auto& rule : rules) {
+      if (!priorities.insert(rule.priority).second) {
+        s->err.message = fmt::format("Found duplicate priority {}.", rule.priority);
+        return -EINVAL;
+      }
+
       rgw_sync_bucket_pipes pipe;
       bool enabled;
       int r = rule.to_sync_policy_pipe(s, driver, &pipe, &enabled);