Retry policy to avoid Thundering Herd Problem


Im configuring the activities retry policy and im a lil bit curious about the Thundering Herd Problem
There is a backoff coefficient but theres no jitter configuration to spread the retries and avoid idle times.

I already have a tasklist configuration to avoid flood but this random coefficient would be interesting to have.
do i have to include this jitter coefficient or temporal comes with this option under the trunk?


Activities are not invoked directly but through task queues. So their invocation is inherently flow controlled as worker get activity tasks only when they have spare capacity. So far we never experienced thundering herd problem in production.