helium/gateway-rs

Queued poc report submission

jeffgrunewald opened this issue · 5 comments

  • Gateway should enqueue beacon and witness reports
  • Beacon reporting should be retried on failure up to 5 times every 5 minutes to the ingest oracle
    • How big should the queue of reports be bounded?
    • Should the retry interval be static (ex: every 10 sec) or backing off (ex: doubling)?

@madninja assuming this is still desirable, can you provide some clarifications on what the retry parameters should be for report submission? static wait in between attempts or exponential backoff up to N attempts?

I wouldn't over complicate it.. start with something simple like try 5 times every 5 minutes to deliver a beacon/witness report

Sure but how many times do we cycle? Up until the next scheduled beacon, or just keep stacking beacons until the queue is empty? What bound should we have on the queue of any?

Sure but how many times do we cycle? Up until the next scheduled beacon, or just keep stacking beacons until the queue is empty? What bound should we have on the queue of any?

Like with normal lora packets we can have the poc (fifo) queue len be a setting. Start with like 5.

With first to witness - HIP 83 queuing witness reports is not useful since any delay at all is going to get you of the list. I'm closing this for now