RetryConfig seems to be ignored on `@tasks_fn.on_task_dispatched`

[sjudd]

I created a task queue function and configured the try config using the decorator:

@tasks_fn.on_task_dispatched(
    secrets=["SOME_SECRET"],
    retry_config=RetryConfig(max_attempts=5, min_backoff_seconds=60),
    rate_limits=RateLimits(max_concurrent_dispatches=1),
)

But when the body of the function throws an Error, I see 15 retries spaced about 10 seconds apart:

I'd expect to see only 5 retries. I'd expect each one to be at least a minute apart.

I'd also expect to see some kind of exponential backoff behavior, but instead the time between retries seems fixed

[exaby73]

Hey @sjudd. Could you give me the logs for one of these errors?

[sjudd]

@exaby73 sorry i missed this. Can you clarify what you're looking for? The error is an incorrectly formatted request to a graphql endpoint.

[exaby73]

Can you expand the error and paste one here with any private information redacted?

[sjudd]

I've swapped the retry config, and the issue persists (retries every 10 seconds):

    retry_config=RetryConfig(
        max_attempts=40,
        min_backoff_seconds=60,
        max_backoff_seconds=16 * 60,
        max_retry_seconds=4 * 60 * 60,
    ),