How to identify the offiline operators if 1/2 operators are malicious in a cluster

[TinyVege]

for example, Distributed Validator A is handled by operator1,2,3,4.
We assume operators1,2 are honest while operators 3-4 is malicious and pretend to be online, so how can the ssv identify exactly the malicous operators 3-4.

[GalRogozinski]

See BFT Forensics

[TinyVege]

See BFT Forensics

I have read it. It only identifies the double vote misbehavior, can not solve the liveness vioaltion right?

My question is that operators3-4 don't send any protocol(validator duty) messages to the dvt cluster and discard the protocol messages of operators1-2. How to prove the operators3-4 are the faultys ?

[MatheusFranco99]

Yes, BFT Forensics can't solve the protocol. On a theoretical point of view, any BFT can only guarantee its properties (agreement, termination, etc) if at maximum $f$ nodes are malicious.

We don't have BFT Forensics yet, but multiple strategies could be added, such as idleness, invalid messages and so on.

[TinyVege]

I'm also interested in evaluating the performance of operators in ssv (shown bolow)

https://docs.ssv.network/learn/performance

https://docs.ssv.network/learn/performance/statuses

Now assume, there are four operator1, operator2, operator3, operator4. '' implies the Byzantine.
And operaotors1-4 only perform the validatorX.
Operators1-3 do the protocol while the operator4 is offline --> The validatorX does not miss any duty.

⚠⚠⚠
So the result is that the performance and status of operator1-4 are the same : 100% and active
Right ? 👀👀👀
@MatheusFranco99 @GalRogozinski

[GalRogozinski]

Your last question is best to be asked on our Discord server
https://discord.com/invite/ssvnetworkofficial

this repo is used for discussing issues of the core protocol itself

[TinyVege]

ok ,thanks