bug: half the committee members casting ballots is not a quorum

[Chris-Hibbert]

closes: #10274

Description

Half the EC were able to pass a motion, when a majority should have been required.

Security Considerations

Ordinary vote counting issue.

Scaling Considerations

N/A

Documentation Considerations

None

Testing Considerations

Added tests for committee.

Upgrade Considerations

The change is in the committee code. This needs to be merged before the proposal in #10164 runs. It already installs new committee code.

[cloudflare-workers-and-pages]

Deploying agoric-sdk with    Cloudflare Pages

Latest commit:	b1a98fa
Status:	✅  Deploy successful!
Preview URL:	https://bd0364b8.agoric-sdk.pages.dev
Branch Preview URL:	https://10274-majorityvote.agoric-sdk.pages.dev

View logs

[dckc]

I suggest trying force:integration. I think that will show that a3p has a 2 member EC and a3p tests rely on being able to carry a question by casting just one vote.

That is: this is a breaking change, and I'm not sure how much integration testing is in order to mitigate surprises.

Mainnet likewise has an ever number of EC members. I believe they rely on being able to make decisions based on 3 votes. I suppose that changes as of #10164 , so that's OK.

I think devnet and emerynet have 3 EC members, so they wouldn't be affected by this change.

[Chris-Hibbert]

the force:integration tests passed.

I think that will show that a3p has a 2 member EC and a3p tests rely on being able to carry a question by casting just one vote.

Can you check your expectations? Does that test exist? Why is it still able to pass?

[dckc]

I think that will show that a3p has a 2 member EC and a3p tests rely on being able to carry a question by casting just one vote.

Can you check your expectations? Does that test exist? Why is it still able to pass?

Looks like it's in a PR that isn't merged yet.

#10241 (comment)

the force:integration tests passed.

That suggests that we could / should merge this one, but I think that would add work to #10241 , and #10241 closes several issues that are higher priority than #10274 . So I'm inclined to wait to merge this until #10241 lands.

Maybe we should increase the EC size in a3p to 3 in #10241 , making it unaffected by this change?

[Chris-Hibbert]

That suggests that we could / should merge this one, but I think that would add work to #10241 , and #10241 closes several issues that are higher priority than #10274 . So I'm inclined to wait to merge this until #10241 lands.

I'm fine with waiting, and leaving the burden here.

Maybe we should increase the EC size in a3p to 3 in #10241 , making it unaffected by this change?

If we increase the EC size to 3, the quorum will be 2, and at least two voters will have to vote, so a3p will be affected either way. I think it's correct for the committee to enforce that more than half the voters have to vote to get a quorum, and there have to be more votes in favor than against to pass a motion.

[Chris-Hibbert]

rebased pose 10241. Let's see how tests do.

[Chris-Hibbert]

@dckc, PTAL. Tests pass.

[dckc]

We're using 3-member committees in a3p and devnet configs now. So this shouldn't cause problems.