Don't commit failed out-dated votes

[sakridge]

Problem

Outdated votes can be bloated on-chain. Since votes arrive out-of-order, then older ones can be applied after later ones and also for different branches.

Summary of Changes

Don't commit failed votes which fail in the way of VoteTooOld.

Fixes #

[codecov]

Codecov Report

Merging #20850 (f4d7211) into master (140a5f6) will decrease coverage by 0.0%.
The diff coverage is 64.0%.

@@            Coverage Diff            @@
##           master   #20850     +/-   ##
=========================================
- Coverage    81.5%    81.5%   -0.1%     
=========================================
  Files         499      499             
  Lines      140278   140296     +18     
=========================================
- Hits       114403   114368     -35     
- Misses      25875    25928     +53     

[carllin]

Nice, so once this goes in, we can then increase the range of leaders in the vote_listener

[tao-stones]

The PR I had before, #19400, does similar thing, except it drops too_old vote before bank.load_and_execute_transactions(...), might be bit more efficient than doing so after bank load/execute.
However, that PR probably needs a bit more work.

[carllin]

The PR I had before, #19400, does similar thing, except it drops too_old vote before bank.load_and_execute_transactions(...), might be bit more efficient than doing so after bank load/execute. However, that PR probably needs a bit more work.

Yeah if we're going to be parsing through hundreds of thousands of votes, it might be preferable to do this before the load_and_execute_transactions, I'll take another pass at this.

[carllin]

Although each InstructionError here incurs extra account locking + copying, I think this change is a lot simpler, so might be safer to check in first.

We might want to bench how long it takes to continually read out vote account + SlotHashes sysvar (this later one is probably bigger) for a few hundred thousand votes.

The failure rate can also be mitigated if we do some more smart processing from cluster_info_vote_listener before we send, I'll see if I can spin that up

[carllin]

Other side of cluster_info_vote_listener I'm working on: #20873

[tao-stones]

Although each InstructionError here incurs extra account locking + copying, I think this change is a lot simpler, so might be safer to check in first.

Agree, this is much simpler.

[tao-stones]

lgtm

Pull request has been modified.

[stale]

This pull request has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs.

[stale]

This stale pull request has been automatically closed. Thank you for your contributions.