Optimize backfill sync to efficiently use reqresp fetched block data #3669
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…p sync(block/range)
g11tech
requested review from
dapplion,
mpetrunic,
twoeths and
wemeetagain
as code owners
|
Code Climate has analyzed commit 7928f92 and detected 0 issues on this pull request. View more on Code Climate. |
Performance Report✔️ no performance regression detected Full benchmark results
|
Codecov Report
@@ Coverage Diff @@
## master #3669 +/- ##
==========================================
+ Coverage 37.13% 37.39% +0.25%
==========================================
Files 321 322 +1
Lines 8706 8796 +90
Branches 1350 1369 +19
==========================================
+ Hits 3233 3289 +56
- Misses 5330 5365 +35
+ Partials 143 142 -1 |
twoeths
reviewed
Jan 27, 2022
| const block = signedBlock.message as TreeBacked<allForks.BeaconBlock>; | ||
| return { | ||
| key: block.slot, | ||
| value: signedBlock.serialize(), |
There was a problem hiding this comment.
what I mean in the issue is that we return both the binary and the signed block in beaconBlocksByRange and beaconBlocksByRoot. So instead of returning Promise<allForks.SignedBeaconBlock[]> in network req/resp api, return something like
type CachedBytes<T> = {bytes: Uint8Array} & T;
async beaconBlocksByRange(
peerId: PeerId,
request: phase0.BeaconBlocksByRangeRequest
): Promise<CachedBytes<allForks.SignedBeaconBlock>[]>and consume that cached bytes without having to call serialize() again in BackFill sync and similar places
@dapplion do we have this type in ssz v2?
There was a problem hiding this comment.
yes, I was wondering about the same 🙂
This format should currently save on hashTreeRoot calcs which would amount for 9% of hashTreeRoot in verifiBlockSequence (#3657 (comment)), and may be some 2-3% on the serialize.
CachedBytes<allForks.SignedBeaconBlock> would be awesome ❤️ awaiting @dapplion 's input regarding ssz v2.
There was a problem hiding this comment.
Right now, ssz v2 only (optionally) caches the hashTreeRoot (of non-tree-backed values); doesn't cache the serialized bytes.
There was a problem hiding this comment.
There was a problem hiding this comment.
added caching capability as that was discussed to be out of scope for ssz v2 👍
g11tech
commented
Feb 1, 2022
| return (typeTree.createTreeBackedFromBytes(bytes) as unknown) as T; | ||
| const treeBacked = typeTree.createTreeBackedFromBytes(bytes) as unknown; | ||
| if (options?.cacheBytes) { | ||
| return (new Proxy({treeBacked, bytes}, cachedTreeBackedProxyHandler) as unknown) as T; |
There was a problem hiding this comment.
Needed another proxy as tree's proxy wasn't letting assign bytes without going through the tree setters. May be post ssz v2 merge, it could be removed.
g11tech
changed the title
|
Closing for now since the network code has diverged. This optimization is good and should definitely be included on a future review of backfill sync |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Motivation
As @tuyennhv pointed out after doing profiler runs, backfill sync could be optimized to use treebacked data from the sync range/block by root.
This PR changes the following:
--sync.backfillBatchSizeto speciify the batch size for backfill syncDescription
Closes #3657