p2p: Fill reconciliation sets (Erlay) attempt 2 #30116
Conversation
|
The following sections might be updated with supplementary metadata relevant to reviewers and maintainers. Code CoverageFor detailed information about the code coverage, see the test coverage report. ReviewsSee the guideline for information on the review process. ConflictsReviewers, this pull request conflicts with the following ones:
If you consider this pull request important, please also help to review the conflicting pull requests. Ideally, start with the one that should be merged first. |
|
I've talked to @naumenkogs about picking this up and he was happy about it. I'm happy to close this if he changes his mind. |
sr-gi
changed the title
sr-gi
force-pushed
the
2023-11-erlay2.1
branch
3 times, most recently
from
047f61c
to
bc8e7d7
Compare
sr-gi
force-pushed
the
2023-11-erlay2.1
branch
2 times, most recently
from
d4b9fc6
to
935fc82
Compare
|
I've slightly extended the approach adding 3 commits to deal with short id collisions, which were not taken into account. Some of this may be squashable, I've added them separately for now so they are easy to diff/review. This would be missing an additional commit/amend to deal with #28765 (comment), which I overlooked when addressing the outstanding comments |
|
🚧 At least one of the CI tasks failed. Make sure to run all tests locally, according to the Possibly this is due to a silent merge conflict (the changes in this pull request being Leave a comment here, if you need help tracking down a confusing failure. |
sr-gi
force-pushed
the
2023-11-erlay2.1
branch
5 times, most recently
from
e3b5988
to
7c94dae
Compare
`m_is_inbound` cannot be changed throughout the life of a `Peer`. However, we are currently storing it in `CNodeState`, which requires locking `cs_main` in order to access it. This can be moved to the outside scope and only require `m_peer_mutex`. This is a refactor in preparation for Erlay reworks.
These comments became irrelevant in one of the previous code changes. They simply don't make sense anymore.
They will be used later on.
Transactions eligible for reconciliation are added to the reconciliation sets. For the remaining txs, low-fanout is used. Co-authored-by: Martin Zumsande <mzumsande@gmail.com> Co-authored-by: Pieter Wuille <pieter.wuille@gmail.com>
|
I added two more commits, moving the fanout/reconciling logic to I'm moving this to draft for now until I clean it a bit, plus get some feedback on the approach |
|
🚧 At least one of the CI tasks failed. Make sure to run all tests locally, according to the Possibly this is due to a silent merge conflict (the changes in this pull request being Leave a comment here, if you need help tracking down a confusing failure. |
sr-gi
force-pushed
the
2023-11-erlay2.1
branch
2 times, most recently
from
fa5df88
to
edb0072
Compare
It helps to avoid recomputing every time we consider a transaction for fanout/reconciliation. Co-authored-by: Sergi Delgado Segura <sergi.delgado.s@gmail.com>
… short ids to wtxids
If a wtxid to be added to a peer's recon set has a shot id collisions (a previously added wtxid maps to the same short id), both transaction should be fanout, given our peer may have added the opposite transaction to our recon set, and reconciliation will fail. Moreover, all descendants of the previously added transaction that were in the recon set should also be removed and fanout.
…stors If a transaction to be reconciled has in mempool ancestors, we need to be consistent with the way in which we announce this transaction (either by fanning out or reconciling all the ancestor set). In order for Erlay to work best, a minimum amount of fanout is required. Therefore, we cannot simply reconcile with everyone in this case either. Find the subset of peers that have the least ancestors to be reconciled and send all of them via fanout
This is a re-attempt of #28765
The main differences from it are:
IsFanoutTargethas been updated to reflect what the algorithm is doing (not how it is doing it)hash_keyis seeded inIsFanoutTargethas changed (fromm_k0towtxid.ToUint256()). This is to prevent usingm_k0for something it is not intended to, given what data we feed to the randomizer should not matterdestinations/targethas been renamed toninShouldFanoutTo/IsFanoutTargetAlso, the following things have been added/changed to the approach:
RelayTransactionThe differences can be easily checked via
git range-diff a14dfd9c873d1e196252c77ad7a8b32bd21b6f6d...ac5dc0aa512e78a1cc90e0ab182eedaec68b0444Erlay Project Tracking: #28646