Weird Apple Push Notification behaviour when the receiving iPhone is turned off

I'm seeing some very strange behaviour from the Apple Push Notification Servers when the recipient iPhone is off. Here is my scenario:

-Send push notification A to Apple. Within a few seconds a push notification popup gets displayed as expected on the iPhone.

-Send blank notification to Apple to cancel previous one (the previous notification is pointless after about 10 seconds, that's why I want to get rid of it). Nothing displayed on the iPhone.

-Turn OFF iPhone completely (not asleep, it is powered down).

-Send push notification B to Apple. Wait 10 seconds.

-Send blank notification to开发者_如何学C Apple to cancel previous one. Wait 10 seconds.

-Send push notification C to Apple. Wait 10 seconds.

-Send blank notification to Apple to cancel previous one. Wait 30 seconds.

-Turn ON iPhone.

-After about 60 seconds a push notification popup is displayed for notification B on the iPhone.

-Notification C never seems to arrive.

This is very strange! From reading the Apple docs I was expecting only the latest push notification to be sent. I was hoping my blank notification would be sent, I certainly wasn't expecting the oldest unsent push notification to be sent!

The Apple docs say:

Apple Push Notification Service includes a default Quality of Service (QoS) component that performs a store-and-forward function. If APNS attempts to deliver a notification but the device is offline, the QoS stores the notification. It retains only one notification per application on a device: the last notification received from a provider for that application. When the offline device later reconnects, the QoS forwards the stored notification to the device. The QoS retains a notification for a limited period before deleting it.

Has anybody seen this behaviour? Am I just hitting some sort of timing window bug? What should happen?

Updates:

-If I turn the phone off and wait 5 to 15 minutes before sending any push notifications then this problem doesn't occur. In this case when I turn the phone on I don't see any notification popup, although I'm not sure if this is a result of Apple dropping the notification, or their 'queue' working correctly (i.e. holding the newest blank notification instead of the first one with the popup).

-I will investigate further by putting an APNsLogging.mobileconfig onto the iPhone to see what notifications it got.

-Turning wifi off doesn't seem to change the results.

-I have raised a bug report with Apple for this scenario.

You may want to check for this behavior across both the cellular and WiFi networks. There's a lot of strange behavior when the phone is on WiFi, especially if there are multiple NAT router involved, i.e. in a large corporation where there's a main router and per-floor WiFi routers, or in a home where you have multiple routers used to extend the range. But on cell it's been pretty solid.

Also, the 10-second cancellation delay may be cutting it too close. They don't guarantee timely delivery and I've gotten lags of as much as 3 minutes on the production server after queuing off a push request. You may want to plan for system congestion.

Either way, it sounds like it might be worthy of a bugreporter report.

For the record, I raised this as bug ID #7349660 with Apple on 29th Oct (https://bugreport.apple.com), then gave them the additional diagnostics they wanted on 30th Oct.

No response from Apple since then, so I am assuming this is probably just low on their priority list, which is fair enough as it is quite a small timing window where the problem can occur (which I didn't realize when I first opened this question).

In apple's document it is said that it can cache the push notification up to 30 days.so u may get push once you switch on your iphone (provided you are having interconnection)

How soon was C sent after B? It seems like it may be a bug where there's some kind of timeout server side, and they accidentally keep B if C was sent too soon after...

If you have a good test example (and this seems like a good test) I would file a Radar report on this.