staking: add manual_slash extrinsic
#7805
Conversation
ordian
added
T1-FRAME
|
All GitHub workflows were cancelled due to failure one of the required jobs. |
| /// will be applied. | ||
| /// - The slash will be deferred by `SlashDeferDuration` eras before being enacted. | ||
| #[pallet::call_index(33)] | ||
| #[pallet::weight(T::WeightInfo::force_unstake(1))] |
There was a problem hiding this comment.
Looks good. Approving modulo the question I've left and your todo.
My view on the questions in the PR description:
should manual_slash accept session instead of an era when the validator was in the active set?
I can't think about a case when we want to slash at a particular session so I'd say this is good.
should the accepted origin be something more than just root?
Maybe we might want to apply a manual slash from AH via XCM, but I wouldn't worry about this right now. I think it's good as it is.
should I adapt this PR also against #6996?
@Ank4n and @kianenigma are doing a lot of changes to the staking pallet here which will be merged in master. I think having your PR merged in master is enough.
| @@ -2639,5 +2639,63 @@ pub mod pallet { | |||
|
|
|||
| Ok(()) | |||
| } | |||
|
|
|||
| /// This function allows governance to manually slash of a validator and is a | |||
There was a problem hiding this comment.
| /// This function allows governance to manually slash of a validator and is a | |
| /// This function allows governance to manually slash a validator and is a |
|
|
||
| // Check era is valid | ||
| let current_era = CurrentEra::<T>::get().ok_or(Error::<T>::InvalidEraToReward)?; | ||
| let history_depth = T::HistoryDepth::get(); |
There was a problem hiding this comment.
Is there a point to check if validator_stash was in the active set during era or this is performed later when handling the offence?
This PR adds a convenience extrinsic
manual_slashfor the governance to slash a validator manually.Changes
on_offenceimplementation for the Staking pallet accepts a slice ofOffenceDetailsincluding the full validator exposure, however, it simply ignores that part. I've extracted the functionality into an inherenton_offencemethod that takesOffenceDetailswithout the full exposure and this is called directly inmanual_slashmanual_slashcreates an offence for a validator with a given slash percentangeQuestions
manual_slashaccept session instead of an era when the validator was in the active set? staking thinks in terms of eras and we can check out of bounds this way, which is why it was chosen for this implementation, but if there are arguments against, happy to change to session index