Why this many Elders? I don't grok the Seven

This is in the CITweaking branch (Continuous Integration Tweaking) so I’d say it’s testing something out related to the continuous integration (ie the automated testing that runs on github etc) and isn’t related to potential changes in network consensus / security. Just a guess though.

Why 5 instead of 7 for CI? Maybe it’s so the tests run faster? Maybe it’s so splits take less effort to induce? Could be a bunch of different reasons, I’m not sure exactly.


Correct @mav , it’s me poking at getting CI on its feet once again. CI testing is something else to the actual network.

In that PR I’ve just made the elder count easily customisable, so we can lower it on CI.

Why? At the mo DKG is still a pain in the initial section set up. CI machines are not powerful (locally 7 elders and 45 nodes is fine for me eg). So I was wondering how reducing it affects us there. Tests run and pass. So that’s good. (Better than constant fails which tell us nothing of new code being added).

Do we want to test CI w/ 5 elders? Ideally not. But until github unlocks more powerful machines (coming Q1 2022 apparently) (or initial DKG is licked… which it may be very shortly :crossed_fingers:, though that’s another story ) it’s something worth considering…

There is no byzantine Q on CI at the moment. So in theory 5 or 7 shouldn’t make any difference in that respect. But it does allow us to run less nodes and get more splits faster on these machines. Though there may be some other reasons not to go down that route… But as I say, as soon as we can reliably run w/ 7 elders on those machines, then we’ll be donig that.