Management and logs
Seeing what was delivered, what failed, and retrying.
Retries
A delivery that does not answer 2xx is retried eight times over about two days and twenty hours. An endpoint down for a weekend still gets everything.
| Attempt | Wait before it |
|---|---|
1 | immediate |
2 | 1 minute |
3 | 5 minutes |
4 | 30 minutes |
5 | 2 hours |
6 | 8 hours |
7 and 8 | 24 hours |
When the endpoint stays down
After 15 consecutive failures the endpoint is deactivated, and an SMS goes to the alert number if you set one. Events do not stop being recorded: for seven days they are kept aside, and reactivating the endpoint delivers the backlog in order. A fault on your side costs delay, never lost events.
Delivery is at-least-once
If we crash between your 2xx and recording it, the same event is delivered again. Deduplicate by the X-DroidSender-Delivery header, which is unique per delivery attempt.
Logs
- Each delivery records the status returned and the response time
- Logs are kept for 30 days
- Test now sends a sample event to your endpoint