This is largely a reimplementation using boost::multi_index_container.
All the same public methods are available. It has an index by outpoint,
per-peer tracking, peer worksets, etc.
A few differences:
- Limits have changed: instead of a global limit of 100 unique orphans,
we have a maximum number of announcements (which can include duplicate
orphans) and a global memory limit which scales with the number of
peers.
- The maximum announcements limit is 100 to match the original limit,
but this is actually a stricter limit because the announcement count
is not de-duplicated.
- Eviction strategy: when global limits are reached, a per-peer limit
comes into play. While limits are exceeded, we choose the peer whose
“DoS score” (max usage / limit ratio for announcements and memory
limits) is highest and evict announcements by entry time, sorting
non-reconsiderable ones before reconsiderable ones. Since announcements
are unique by (wtxid, peer), as long as 1 announcement remains for a
transaction, it remains in the orphanage.
- This eviction strategy means no peer can influence the eviction of
another peer’s orphans.
- Also, since global limits are a multiple of per-peer limits, as long
as a peer does not exceed its limits, its orphans are protected from
eviction.
- Orphans no longer expire, since older announcements are generally
removed before newer ones.
- GetChildrenFromSamePeer returns the transactions from newest to
oldest.
Co-authored-by: Pieter Wuille <pieter@wuille.net>
Move towards a model where TxOrphanage is initialized with limits that
it remembers throughout its lifetime.
Remove the param. Limiting by number of unique orphans will be removed
in a later commit.
Now that -maxorphantx is gone, this does not change the node behavior.
The parameter is only used in tests.
Deduplicate the logic of adding the parents as announcements to
txrequest. The function can return a bool (indicating whether we're
attempting orphan resolution) instead of the delay.
Now that we track all announcers of an orphan, it's not helpful to
consider an orphan provided by a peer that didn't send us this parent.
It can only hurt our chances of finding the right orphan when there are
multiple candidates.
Adapt the 2 tests in p2p_opportunistic_1p1c.py that looked at 1p1c
packages from different peers. Instead of checking that the right peer
is punished, we now check that the package is not submitted. We can't
use the functional test to see that the package was not considered
because the behavior is indistinguishable (except for the logs).
This means we no longer return parents we already have in the
m_unique_parents result from MempoolRejectedTx.
We need to separate the loop that checks AlreadyHave parents from the
loop that adds parents as announcements, because we may do the latter
loop multiple times for different peers.
Avoid the fuzzer situation where:
1. Orphanage has 2 transactions with the same txid, one with witness,
one without witness.
2. The transaction with witness is found to have
`TX_INPUTS_NOT_STANDARD` error. The txid is added to recent rejects
filter, and the tx with witness is deleted from orphanage.
3. A low feerate parent is found. Find1P1CPackage finds the transaction
with no witness in orphanage, and returns the package.
4. net_processing has just been handed a package in which the child is
already in recent rejects.
This is a slight behavior change: if a transaction is in both
reconsiderable rejects and AlreadyHaveTx in another way, we don't try to
return a 1p1c package. This is the correct thing to do, as we don't want
to reconsider transactions that have multiple things wrong with them.
For example, if a transaction is low feerate, and then later found to
have a bad signature, we shouldn't try it again in a package.
This will become necessary in later commits that query mempool. We also
introduce the TxDownloadOptions in this commit to make the later diff
easier to review.
This module is going to be responsible for managing everything related
to transaction download, including txrequest, orphan transactions and
package relay. It will be responsible for managing usage of the
TxOrphanage and instructing PeerManager:
- what tx or package-related messages to send to which peer
- whether a tx or package-related message is allowed or useful
- what transactions are available to try accepting to mempool
Future commits will consolidate the interface and re-delegate
interactions from PeerManager to TxDownloadManager.
