p2p protocol should be optimized to not requests blocks that don't exist when syncing from genesis

[matthewdarwin]

trafficstars

Matthew Darwin - EOS Nation, [21.09.21 23:24] When starting nodeos from genesis and some of the peers it connects to have all the blocks and some of the peers it connects to don't have all the blocks (started from snapshot), nodeos takes a long time to sync because it keeps requesting blocks from nodes which don't have them. Other than figuring out (by reading the logs) which peers are missing the blocks (and then removing the connection from nodeos), is there a way to speed up nodeos to have it realize which peers don't have blocks and stop asking them? Would this be a good enhancement request? Or the p2p protocol can't handle this? @heifner

Kevin Heifner, [22.09.21 08:06] [In reply to Matthew Darwin - EOS Nation] Yes, this would be a good enhancement. Can you put that into a github issue please.

EOSUSA Michael, [22.09.21 08:18] didnt u say u had already addressed the unlinkable blocks in another fix or would that fall into this feature?

Kevin Heifner, [22.09.21 08:21] The fix will help this, but it does still request blocks from nodes that it has already been told does not have the blocks ( it is optimistic that it will have the new range of blocks). We could add communication of not only the max but the min of blocks.