For example, old versions of vipnode that are incompatible with new versions could yield errors on peering requests. We should penalize them lightly so that they're not super-recurring.
Ideas for penalizing:
- Reduce priority relative to something? Maybe probabilistic? (E.g. skip 50% of the time, incremented to 100% with each failure? Decremented over time? Too complicated?)
- Put a "timeout" on them so that they're not considered for 10min? 60min?
Penalty should be reset on reconnect? Or should penalty slowly decrease on each success?
For example, old versions of vipnode that are incompatible with new versions could yield errors on peering requests. We should penalize them lightly so that they're not super-recurring.
Ideas for penalizing:
Penalty should be reset on reconnect? Or should penalty slowly decrease on each success?