Round 51679 count inconsistency?

[ldeffenb]

Goerli round 51679 shows 5 commits but 9 reveals? The has summary seems to support the 9 as well. The details of this round need to be examined and the discrepancy explained. Possibly a re-org in the middle of the round?

08:33:48 51679(81) 5-9 7^2+2^1=1 675d2d8f3... ^2 +1.58P

[mfw78]

This sounds like a decent candidate for the invariant num_reveals <= num_commits

[ldeffenb]

Ok, this looks like it may be difficult to track down as re-executing just round 51679 (blocks 7855208 to 7855359 doesn't show any game participation in that round. Strange, and seems further to point to a possible re-org that messed things up. Maybe it's time to change the block skip detector to an invariant because I don't know what the block subscription actually does in a re-org.

monSI/src/chain/sync.ts

Line 302 in 78c9237

Logging.showError(

But then again, I really don't want the monitor stopping for the user, so maybe not do the invariant just yet, but keep our eye on the behaviors.

[ldeffenb]

Did additional replay testing and cannot reproduce the error display on that round. We'll keep this issue open just in case such a strangeness is observed again in some future round.

From the re-run:
08:33:48 51679(81) 5-5 4^2+1^1=1 675d2d8f3... ^2 +1.58P