Currently, the HTTP workers do not wait between retries. Retrying calls without waiting can produce a heavy load on the Remote RMM, also, retrying too fast will not help to get a result in a momentary outage.
I suggest considering exponential backoff for worker retries where the worker periodically retries a failed request with increasing delays between requests.
Benefits:
The increase in the wait time gives chance to get a result from the remote RMB.
The RMB server won't ever get flooded because of the backoff.
Currently, the HTTP workers do not wait between retries. Retrying calls without waiting can produce a heavy load on the Remote RMM, also, retrying too fast will not help to get a result in a momentary outage.
I suggest considering exponential backoff for worker retries where the worker periodically retries a failed request with increasing delays between requests.
Benefits: