BEEFY: Error when using fast sync and BEEFY genesis is 1

[serban300]

When using fast sync or warp sync and BEEFY genesis is 1 we get a Could not find initial validator set error, even if the block sync finished and we are running with --pruning x where x > chain length

[serban300]

This is the function that generates the failure:

First it tries to call runtime.runtime_api().validator_set(at_header.hash()) and then, if this fails, it tries to search for autority set changes in the header log: worker::find_authorities_change::<B>(&header).

The problem is that:

• from what I understand, when using fast/warp sync the state for old blocks is not applied, even if we run the node with --pruning x where x > chain length. So we don't have the state at block 1 => runtime.runtime_api().validator_set(at_header.hash()) will fail.
• also at block 1 no authority set change was logged in the header log. Block 1 contains only the initial authority set in its state => worker::find_authorities_change::<B>(&header) will fail.

[acatangiu]

We should fix this by adding a new ConsensusLog::ConsensusReset(active_validator_set), that way we can get all required information from headers - which we need in case fast/warp sync where state is not available.

There is a pre-monorepo PR for this - you can port that and drive it to completion.

In the meantime, to get warp-sync zombienet tests working, you should set BEEFY genesis on a session boundary block instead of block 1. Or even better, do a quick fix to the BEEFY worker to get active set from state of chain genesis block (block zero). AFAIK state is always available for block zero.