RetryxRetry Calculator
Model an exponential backoff schedule with jitter and see the exact retry times. Enter how long your receiver might be down to check whether the events would survive.
Survives this outage. Typically delivered on attempt 8, about 47.6m after the first failure (between 31.8m and 1.1h with jitter).
| Attempt | Wait before | Time since first failure | With jitter |
|---|---|---|---|
| 1 (first try) | – | 0s | – |
| 2 | 30s | 30s | 15s to 30s |
| 3 | 1m | 1.5m | 45s to 1.5m |
| 4 | 2m | 3.5m | 1.8m to 3.5m |
| 5 | 4m | 7.5m | 3.8m to 7.5m |
| 6 | 8m | 15.5m | 7.8m to 15.5m |
| 7 | 16m | 31.5m | 15.8m to 31.5m |
| 8 | 32m | 1.1h | 31.8m to 1.1h |
Defaults match Retryx’s default policy: 8 attempts, 30 s first delay, ×2, capped at 6 hours, with 50–100% jitter. “Typical” uses the middle of the jitter range.
Exponential backoff for webhooks
When a delivery fails, retrying immediately usually fails again and can make an outage worse. Exponential backoff waits longer after each failure: for example 30 seconds, then 1 minute, then 2, 4, 8 and so on, up to a maximum delay. Adding jitter randomises each wait so thousands of failed deliveries do not all return at the same instant.
The important question is the total window: how long after the first failure does the system keep trying? If your receiver can be down for an hour and the schedule gives up after ten minutes, those events are gone. The calculator uses the same formula as Retryx: the wait after failed attempt N is the smaller of the maximum delay and base × factorN−1, with 50–100% jitter. Read more in delivery and retries.
How to use it
- Enter the first retry delay, the backoff factor and the maximum delay in seconds.
- Set the total number of attempts (the first try counts as attempt 1).
- Enter how long your receiver might be unavailable, in minutes.
- Read the verdict and the schedule table, and adjust until it covers your worst realistic outage.
Frequently asked questions
- What are good defaults?
- Eight attempts, a 30 second first delay, a factor of 2 and a 6 hour cap, which are Retryx’s defaults, cover roughly half a day of cumulative waiting.
- What is jitter?
- Random variation added to each delay. The calculator shows the range with 50–100% jitter; “typical” is the middle of that range.
- Does “attempts” include the first try?
- Yes. Attempt 1 is the original delivery, so 8 attempts means 7 retries.
- Which failures should be retried?
- Network errors, timeouts, 408, 425, 429 and 5xx responses. Other 4xx responses mean the request itself is wrong and retrying will not help.
- What if the schedule runs out?
- The event is dead-lettered. In Retryx it is kept and you can replay it once the receiver is healthy again.