Slashing should reduce registered amount that is bonded

[Tbaut]

The account 5F7BJL6Z4m8RLtK7nXEqqpEqhBbd535Z3CZeYF6ccvaQAY6N has more DOTs bonded than available in free_balance. Also it has no vesting and no reserved balance.

staking.ledger from its controller:

{
    "stash": "5F7BJL6Z4m8RLtK7nXEqqpEqhBbd535Z3CZeYF6ccvaQAY6N",
    "total": "0x00000000000000000023829049fb9940",
    "active": "0x00000000000000000023829049fb9940",
    "unlocking": []
}

which is ~9,9951×10¹⁵

balances.freeBalance: ~9,9946x10¹⁵

[kianenigma]

Interesting. Keeping these two in mind:

• https://github.com/paritytech/substrate/blob/master/srml/staking/src/lib.rs#L579
• https://github.com/paritytech/substrate/blob/master/srml/staking/src/lib.rs#L599

I don't see how this account ended up like this. Is this on Alexander?

[jacogr]

Alex, yes.

Currently the same account is unlocking (aka entered the 2 year wait), however the same thing now shows slightly differently - the total value being unlocked is now > freeBalance.

[jacogr]

This may help as well, only 2 extrinsics on this account - https://polkascan.io/pre/alexander/module/account/5F7BJL6Z4m8RLtK7nXEqqpEqhBbd535Z3CZeYF6ccvaQAY6N

• staking.setKey
• staking.bond with 10000000000000000 (10 DOTs)

The controller is here - https://polkascan.io/pre/alexander/module/account/69edffba4af672942bf6f60dfe839a20abf4556bb28a7c470a1778291326c46c (linked from the bond extrinsic)

[kianenigma]

Great example. and now I feel more serious about this. The locks fo this acc seems quite off to me, or am I missing sth? The first two locks both contains an amount which is almost equal to the total bonded amount.

.

Will further investigate this next week.

[Tbaut]

Comment offline from @kianenigma: the original issue that staking.active>freeBalance could be valid if the account has been slashed. So this is actually ok.
I could confirm it on an account that has been slashed (found on Alex) e.g here the total is also higher than the available.

I'd suggest to open another issue for the locks that are off.

[kianenigma]

@Tbaut any other strange behavior regarding this? Now, reading the discussion again, I think the original offset was simply because of paying fees.

[Tbaut]

It's not about the fee, it's about the registered amount (see this comment) and the new title of this issue.

[kianenigma]

This is fixed already, as we do call update-ledger now with the new amount bonded after the slash.

substrate/frame/staking/src/slashing.rs

Lines 599 to 606 in 9467f56

	slashed_imbalance.subsume(imbalance);
	if !missing.is_zero() {
	// deduct overslash from the reward payout
	*reward_payout = reward_payout.saturating_sub(missing);
	}
	<Module<T>>::update_ledger(&controller, &ledger);