Important clarification: How is the token refresh policy applied in a clustered environment, with multiple hosts generating their own JWT tokens with the same key? (The documentation does state "You can use the same token from multiple provider servers," but it's unclear what happens when those multiple servers have to refresh the token.)
In other words, let's suppose you have two provider servers posting to APNS. Can each of those hosts generate their own JWT token (such that each server will have its own refresh frequency, f, such that 20 min < f < 60 min, starting at different times t1 and t2, such that t1 <> t2, |t2-t1| < 20 min) without throwing the TooManyProviderTokenUpdates exception? Or would one host need to be aware of the token refresh on the other host, to avoid regeneration within 20 minutes? Above, there is a mention of connection becoming unusable, so is the JWT refresh policy applied per connection, per host, or per organization/key (i.e. single token across all providers) on the APNS side? To build a centralized token refresh management service is not trivial (due to race conditions, single-point-of-failure, etc.), so we are trying to avoid that at all costs.
Finally, we ran a few tests in against development APNS such that we regenerated the token for each request. We did not get any exceptions. Does the validation for the token age behave differently in development and/or is it somehow adaptive whereby the validation kicks in after n occurrences?
(There are multiple questions related to this across different forums, including this one; and I think this thread is closed to answer this question. Thank you!)
