Using branch sergio/test-vote-extensions-on-do

• While testing on this (temporary) branch, which merges main (as of last Fri) into feature/abci++vef, we found two metrics-related bugs (one as of 8ba3231ec, the other as of 72451db8b). See Fixed two bugs related to metrics #562.

• When testing with hash 7d8c9d426 of the branch above, we managed to run a 200-node testnet (4 experiments) and gathered the following metrics.

  • I'm also including the results from v0.37.0-alpha.3 last month for comparison.
  • Every time, the first plot is v0.37.0-alpha.3, and the second, sergio/test-vote-extensions-on-do.
  • I won't include all the metrics (I can provide them on demand... they're also on the DO storage for CometBFT). I will only show the most relevant

• All experiments (v0.37.0-alpha.3 vs. sergio/test-vote-extensions-on-do). We chose the third one (UUID=ddf12...) as it has the least ideal plot.

• Average mempool size
  
  

• CPU (note that sergio/test-vote-extensions-on-do is running the demo code that verifies vote extension signatures both at PrepareProposal and at ProcessProposal)

• Block rate

• Transaction rate

CONCLUSION:

• No noticeable differences in metrics w.r.t v0.37.0-alpha.3 (other metrics not shown here have also been compared)
• We are good to merge feature/abci++vef back into main