21:29:21 i think the ghost wedge has finally been figured out: https://github.com/Fountain5405/monerod-memory-research/blob/master/reports/E001-sync-wedge-ISSUE.md 21:33:01 pretty sure this is from stressnet scaling being accellerated? 21:34:41 > I have not checked whether mainnet's block-weight rules permit this pattern. 21:34:43 tell it to check 21:40:31 iirc, the max block size is 2x the short term median (so 2x the median of the last 100 blocks), which, iirc.., the dynamic bss looks back 100 blocks for the largest block, which for empty blocks would be a max of 300kb on master and 625kb on stressnet. so, if those blocks were empry on fcmp, the next 100 blocks cant be more than something like 60-90mb 21:43:50 (but on stressnet, they can be bigger ofc). disclaimer: im going off of memory 22:33:27 It just reminds me of ghost wedge issues in the past where the solution was "pop some blocks and restart" 22:44:49 those were usually from being forked off iirc