Expose AuthorFilter Call enum

[sea212]

Closes #697

We have to set the eligibleCount of the author_slot_filter pallet to 1 to give every collator even chances of collation and to significantly reduce the number of forks and reorgs. To do so, root has to call a dispatchable of the author_slot_filter pallet. Unfortunately the Call enum of the author_slot_filter pallet is not exposed in the runtime, rendering setting this value through governance impossible.

[Chralt98]

You found a temporary fix of disabling the cache, preventing the storage root mismatch because of the known bug. So my first question was, if this PR is still required then. I could imagine, that we would like to use the extrinsic of the pallet_author_slot_filter in the future. So I would lean towards approving.

It might be, that I don't get the full context now, but wouldn't setting the eligibleCount to 1 to only allow one collator to produce blocks (per active slot)? I could imagine, that this would lead to allowing one malicious lucky collator to rule the network with manipulated blocks.

[sea212]

You found a temporary fix of disabling the cache, preventing the storage root mismatch because of the known bug. So my first question was, if this PR is still required then. I could imagine, that we would like to use the extrinsic of the pallet_author_slot_filter in the future. So I would lean towards approving.

It might be, that I don't get the full context now, but wouldn't setting the eligibleCount to 1 to only allow one collator to produce blocks (per active slot)? I could imagine, that this would lead to allowing one malicious lucky collator to rule the network with manipulated blocks.

@Chralt98 Just realized I referenced the wrong issue in this PR. Does #697 (comment) answer your question?

[sea212]

In addition to that, collators cannot craft blocks as they like, they have to adhere to the consensus rules, other nodes will reject invalid state transitions that don't adhere to the consensus rules. A collator can control which transactions are executed though. If the complete active collator set is occupied by a malicious entity, it can censor transactions (just never include them) or even halt the network.

[Chralt98]

Thanks for your explanation. But why does the pallet_author_slot_filter set the default value of eligibleCount to 50? I mean, the number of forks and reorgs in other parachains should be also very high then. As long as the other collators validate the eligible collator which produces a new block, setting eligibleCount to 1 sounds reasonable.

[maltekliemann]

👍