I've been using DeviceActivityMonitor for 2 years, and recently noticed the following issue, starting in iOS 17.5 (another user also reported here).
For a sizable percentage of my users, device activity event thresholds get triggered together. My app sends notifications for every hour of screen time during the DeviceActivitySchedule using event thresholds. Often users will get, for example, the 1, 2, and 3 hour screen time notifications all at the same time.
I have a hypothesis for why this is happening: the system sometimes terminates the app extension for various reasons, one being if the 6MB memory limit is reached. It seems as though the retry policy is to retry the failed threshold at the next event threshold. And if the following threshold also fails, they can pile up until the next one succeeds. I think this is a new retry policy since iOS 17, and I believe this because:
- There used to be a bug where the same threshold was triggered multiple times in a row, indicating that the failed threshold was retried immediately. This bug is no longer around and it's been replaced by the one I am reporting.
- According to my logs, thresholds that get triggered together are also called earlier when they are supposed to be called - but the callback function does not complete. So this indicates that the threshold isn't just called late, but that it is called once and then retried again later.
If anyone could answer the following questions I'd be super grateful:
- Is there ANY way to log when the system terminates the app extension and for what reason? And not just on my own device, but for all our users in production (because it's hard to reproduce this issue, as it only happens for some portion of our users). Maybe some kind of crash report or failure callback that will allow my to ping my server?
- Could anyone at Apple could confirm my hypothesis about the new retry policy causing this issue?
