Consider reducing 'getblocks' limit below 500 #2368

500 blocks is quite a lot of data. I wonder if we shouldn't target a byte size rather than block count, between requests. e.g. 32MB or somesuch.

It's size limited right now because of the buffer targeting.

@laanwj no longer an issue?

"getblocks" is no longer user in headers-first sync. And the results of "getheaders" (2000 at once!) are dealt with differently.

Thanks @sipa

This issue confused me: 'getblocks' returns 'invs', not actual block data. The problem was always in how we handled block invs, not so much getblocks itself. But as @sipa says, that too has been solved with the new mechanism.