I think we should not simply unwrap_or(None) here, at least without some considerations:
The self.get_one_time_key() returns sth like Result<Option<OTK>, Error>, so it should already return Ok(None) when an OTK cannot be retrieved. However, other errors can still happen and should be somehow propagated.

I see two options here:

1. Either we back to await? here - this results in RPC failing when AWS/DynamoDB errors occur, but not when no OTKs are available
2. Or we discard all errors like you did here, so the RPC will succeed with None, but we should either:
   • Introduce here a log displaying the error message. This is more future-proof but might result in duplicated error alerts in our dashboard.
   • Or make sure all possible failures inside self.get_one_time_key() (and inner calls of self.get_one_time_keys()) have their respective error logs. This requires no action now, but increases maintenance cost in the future - we need to remember that errors are discarded without notice at some layer above.

After a brief look, the conditions for option 2b are met, but I'd appreciate double-checking it.

The same applies for the get_keyserver_keys

i think 2a makes sense. the duplication is a minor issue IMO