When we have a phase-2 IBC channel connection, we are forwarding Bitcoin timestamps and proofs for them
To generate the proof, we are switching the context to a past block height in order to retrieve the data for a prior state
Validators that have a high pruning policy (i.e. only keep very few recent blocks), can’t switch to the past state to generate the data, leading to non-determinism
The hotfix involves removing broadcasting of BTC timestamps. We are going to implement a longer term solution in which we do not get historical states for the generation of timestamps for the next testnet. While this is not consensus breaking, this happened only on a recent block, so if we roll back to that block and apply this it consensus won't break for new participants on the ledger.
The hotfix involves removing broadcasting of BTC timestamps. We are going to implement a longer term solution in which we do not get historical states for the generation of timestamps for the next testnet. While this is not consensus breaking, this happened only on a recent block, so if we roll back to that block and apply this it consensus won't break for new participants on the ledger.