eth65 fix

[matkt]

Signed-off-by: Karim TAAM [email protected]

PR description

If we look at the specification GetPoolTransactions can allow you to retrieve up to 256 transactions at one time while in most cases we only do it for 1 transaction.

For each NewPooledTransactionHashesMessage we do a GetPooledTransactionsFromPeerTask which can be problematic if the peer sends us a NewPooledTransactionHashesMessage per transaction in the pool. If we add each time a request in Netty's queue, this can explain the fact that it is very quickly overloaded.

This PR adds a waiting list for NewPooledTransactionHashesMessage in order to group several hashes into a single GetPooledTransactionsFromPeerTask

By default, hashes are put in a 500ms waiting list. This waiting list is created when receiving a first hash from a peer. This duration can be modified thanks to this flag --Xeth65-tx-announced-buffering-period-milliseconds

Test performed

3 days of non-stop synchronization for the version with the fix. I also compared my results with the besu-ohio-mainnet-1 version to check that I do not consume a lot of memory or that I do not lose too often peers

Changelog

• I thought about the changelog and included a changelog update if required.

[RatanRSur]

Feel free to revert any of the cosmetic changes I pushed :)

All the tests and logic LGTM!