Description

The hydra-node only records latest chainState when observing a head transition; can lead to long catchup times in periods of inactivity

It has been reported that the hydra-node always replays the state from the last snapshot's chainPoint; i.e. if no shapshot is signed but the chain progresses, then upon restart, the chain state from that point will be lost, and will have to be caught up again. This can lead to large delays in catching back up, depending on how long you are offline for.

Suggested solution

Update the chainState more frequently than just when the head logic results in a transition.

Alternative

No response

Additional context

@Crypto2099 care to add any other detail?