net_processing: remove mapRelay, relay txs from m_most_recent_block
ClosedPublic
Actions

Authored by PiRK on Jul 15 2025, 12:53.

Details

Reviewers

Fabien

Group Reviewers

Restricted Project

Commits

rABC733f231e8393: net_processing: remove mapRelay, relay txs from m_most_recent_block

Summary

mapRelay (used to relay announced transactions that are no longer in the mempool) doesn't have an absolute memory limit, only an implicit one based on the rate of transaction announcements.
Remove it.

Note that we can't remove a transaction from the mempool with a replacement transaction, as the sendrawtransaction(tx_b) RPC would fail with txn-mempool-conflict, but as of D15279 we can use avalanche to reject a tx from the mempool.

This is a backport of core#27625
Depends on D18366

Test Plan

ninja all check-all

Diff Detail

Repository

rABC Bitcoin ABC

Lint

Lint Not Applicable

Unit

Tests Not Applicable

Event Timeline

PiRK created this revision.Jul 15 2025, 12:53

Herald added a reviewer: Restricted Project. · View Herald TranscriptJul 15 2025, 12:53

PiRK requested review of this revision.Jul 15 2025, 12:53

teamcity edited the summary of this revision. (Show Details)Jul 15 2025, 12:53

lint: sort imports

test/functional/p2p_leak_tx.py
24 ↗	(On Diff #54853)	without this the test takes 41s, with it it runs in 8s

Harbormaster completed remote builds in B33880: Diff 54852.Jul 15 2025, 13:00

Harbormaster completed remote builds in B33881: Diff 54853.Jul 15 2025, 13:07

How does it work during a reorg ?

This revision now requires changes to proceed.Jul 15 2025, 14:19

add a reorg test
The transaction is put back in the mempool so it should be available for request

PiRK added inline comments.Jul 15 2025, 15:56

test/functional/p2p_leak_tx.py
148 ↗	(On Diff #54860)	Note that this still works on my machine in the first iteration, it does not seem affected by the `self.noban_tx_relay = True` (which can affect the speed of announcing / relaying transactions to peers)