Just to clarify our main concern: We understand the documented limitations of APNs 410 responses, including that their timing is intentionally unpredictable and that APNs may return HTTP/200 for tokens that are no longer in active use for an extended period. Our main question is about the temporary glitch you mentioned. On our side, the drop in HTTP/410 “Unregistered” responses has continued since around May 12 and is still ongoing as of May 22. Is there an estimated timeline for when this temporary glitch is expected to be resolved and APNs 410 “Unregistered” response behavior is expected to return to normal?

Hi, Thank you for the clarification. We understand that APNs 410 timing is intentionally unpredictable, and that APNs may return HTTP/200 for tokens that are no longer in active use for an extended period. However, the current issue does not appear to be correcting itself within a short while on our side. We have continued to observe the same behavior since around May 12, and it is still ongoing as of May 22. Across multiple bundle IDs, the volume and rate of HTTP/410 “Unregistered” responses remain significantly lower than the pre-May-12 baseline, while our APNs request volume, request flow, credentials, push token collection, and App Store Connect deletion trends remain stable. Could you clarify the following? Is this temporary glitch still ongoing on the APNs side? Is there an estimated timeline for when APNs 410 “Unregistered” response behavior is expected to return to normal? Should we expect APNs to eventually return 410 for tokens that would have returned 410 during the affected period, or will those tokens potentially continue returning 200 indefinitely? Is this issue affecting all APNs provider API traffic, or only certain teams, bundle IDs, environments, regions, or token cohorts? Is there any recommended mitigation for providers that rely on APNs 410 responses to clean up inactive tokens? Can you confirm whether App Store Connect app deletion metrics and APNs token invalidation / 410 feedback are produced by separate systems and may diverge during this issue? We can provide affected bundle IDs, Team ID, sample request timestamps, and aggregate response-code trends if that helps your investigation. Thanks.