咨询Microsoft Graph API推送通知未获2xx响应时的重试时间间隔

Great question! I’ve worked through similar sync scenarios with Microsoft Graph, so let’s break down the retry and backlog behavior you’re seeing:

• Exponential Backoff for Initial Failed Notifications: When your server doesn’t return a 2xx status code, Graph starts with short retry intervals (a few seconds) that grow exponentially (roughly doubling each time) up to a certain threshold. But if your server stays offline for an extended period, these retries pause after a while—Graph won’t keep hammering an unreachable endpoint indefinitely.
• Backlog Processing Once You’re Online: Once your server comes back up, Graph doesn’t dump all missed notifications immediately. Instead, it batches them and processes the backlog over a controlled window. From real-world experience and official guidance, this window usually falls between 10 to 30 minutes—your 10-15 minute wait is totally typical here. The exact timing can shift a bit based on how many events were missed and overall service load.
• Critical Heads-Up: If your server is unreachable for more than 48 hours, your webhook subscription might get automatically disabled. You’ll need to re-create the subscription in that case to resume notifications.

A quick note: While Microsoft doesn’t publish exact second-by-second retry timings for backlogs, the range you observed is consistent with how Graph handles delayed notification delivery for Office 365 event subscriptions.

内容的提问来源于stack exchange，提问作者Jedi3112