[WIP] stale-balance command: force-sync EigenPods with slashed validators
#122
Conversation
jbrower95
changed the title
stale-balance command -- force-sync EigenPods with slashed validatorsstale-balance command: force-sync EigenPods with slashed validators
| { | ||
| Name: "stale-balance", | ||
| Args: true, | ||
| Usage: "If needed, calls `verifyStaleBalances` to correct the balance on a pod you don't own. This will attempt to (1) conclude any existing checkpoint on the pod, and then (2) invoke verifyStaleBalances() to start another, more up-to-date, checkpoint.", |
There was a problem hiding this comment.
"calls verifyStaleBalances to correct the balance start a checkpoint on a pod you don't own"
| { | ||
| Name: "stale-balance", | ||
| Args: true, | ||
| Usage: "If needed, calls `verifyStaleBalances` to correct the balance on a pod you don't own. This will attempt to (1) conclude any existing checkpoint on the pod, and then (2) invoke verifyStaleBalances() to start another, more up-to-date, checkpoint.", |
There was a problem hiding this comment.
Note that anything surrounded by backticks (eg `verifyStaleBalances` above) gets added as a "variable name"/"target" in the CLI help text. Would recommend not using these unless you're trying to change the help text
Run ./cli -h to see what i'm talking about
| if currentCheckpointTimestamp > 0 { | ||
| // TODO: complete current checkpoint | ||
| fmt.Printf("This eigenpod has an outstanding checkpoint (since %d). You must complete it before continuing.", currentCheckpointTimestamp) | ||
|
|
||
| proofs, err := core.GenerateCheckpointProof(ctx, eigenpodAddress, eth, chainId, beacon) | ||
| core.PanicOnError("failed to generate checkpoint proofs", err) | ||
|
|
||
| txns, err := core.SubmitCheckpointProof(ctx, sender, eigenpodAddress, chainId, proofs, eth, batchSize, noPrompt) | ||
| core.PanicOnError("failed to submit checkpoint proofs", err) | ||
|
|
||
| for i, txn := range txns { | ||
| fmt.Printf("sending txn[%d/%d]: %s (waiting)...", i, len(txns), txn.Hash()) | ||
| bind.WaitMined(ctx, eth, txn) | ||
| } | ||
| } |
There was a problem hiding this comment.
This could be quite expensive - should we add a flag that signals that you want to complete an existing checkpoint if it's there?
There was a problem hiding this comment.
like i'm picturing a worst-case scenario where we run the health checker on Renzo's pods and out of 8k validators one got slashed
and then we auto-complete a checkpoint for them
which, at 15 gwei, is ballpark ~6/7 ETH. Then, we start a checkpoint for them and complete a second one, for another 6/7 ETH 😓
There was a problem hiding this comment.
(i understand that that's what the health checker is for to an extent, but IMO in a case like this it should alert us rather than auto-completing the checkpoint, because there's a possibility here where we burn 14 ETH without thinking about it!)
There was a problem hiding this comment.
submitCheckpointProof has warning messages all over it (see how we pass noPrompt through). We should indicate ahead of time that starting the checkpoint could lead to finishing an expensive one though..
but that's sort of the same problem we face on checkpoints rn in the CLI -- we don't do any cost/gas estimation yet
jbrower95
changed the title
stale-balance command: force-sync EigenPods with slashed validatorsstale-balance command: force-sync EigenPods with slashed validators
|
going to rebase / reopen this. |
stale-balance
If an EigenPod has been slashed, you can use
stale-balanceto start a checkpoint against the EigenPod. We suggest avoiding doing this if the deviated amount is small (<5%), but it is a legal operation within the protocol at any time as a way to sync the beacon state with the execution state, when we know that a beacon slashing has occurred.To use;
./cli stale-balance --podAddress $EIGENPOD_ADDRESS --execNode $NODE_ETH --beaconNode $NODE_BEACON --sender $EIGENPOD_PK --slashedValidatorIndex $SLASHED_VALIDATOR_INDEXThis will prompt to complete any in-flight checkpoints, and then call verifyStaleBalances.