feat(l1): download all headers from new to old as first step of full sync #4812
Conversation
This file contains hidden or 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
fmoletta
changed the title
Lines of code reportTotal lines added: Detailed view |
…te::into_fullsync
fmoletta
changed the title
fmoletta
force-pushed
the
full-sync-download-headers-first
branch
from
fbfca51 to
119c5cf
Compare
mpaulucci
reviewed
Oct 16, 2025
MegaRedHand
reviewed
Oct 16, 2025
| /// Block headers downloaded during fullsync column family: [`u8;_`] => [`Vec<u8>`] | ||
| /// - [`u8;_`] = `block_number.to_le_bytes()` | ||
| /// - [`Vec<u8>`] = `BlockHeaderRLP::from(block.header.clone()).bytes().clone()` | ||
| const CF_FULLSYNC_HEADERS: &str = "fullsync_headers"; |
There was a problem hiding this comment.
I think this could be merged with the main headers table. I opened #4903 to check that.
mpaulucci
reviewed
Oct 16, 2025
MegaRedHand
reviewed
Oct 17, 2025
Comment on lines
+472
to
+473
| if !block_headers.is_empty() | ||
| && are_block_headers_chained(&block_headers, &order) |
There was a problem hiding this comment.
This was a change in behavior from later versions we noticed. Before, empty responses were never forwarded to the user, but somewhere down the line it changed, producing errors that were previously unreachable.
Comment on lines
+357
to
+358
| let first_header = block_headers.first().ok_or(SyncError::NoBlocks)?; | ||
| let last_header = block_headers.last().ok_or(SyncError::NoBlocks)?; |
There was a problem hiding this comment.
These errors are actually unreachable, since request_block_headers_from_hash never returns an empty Vec. I think we should change, in another PR, request_block_headers_from_hash to return a plain Vec instead of an Option<Vec<>>, and interpret an empty Vec the same as the current None.
edg-l
approved these changes
Oct 17, 2025
| // Stores current Snap State | ||
| snap_state: SnapState, | ||
| // Stores fetched headers during a fullsync | ||
| fullsync_headers: HashMap<BlockNumber, BlockHeader>, |
There was a problem hiding this comment.
if this is expected to have less than < 10000~ it might be better to use a btree too here
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
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
In order to properly handle reorgs, we should download headers from the latest announced fcu head until me meet our local chain instead of requesting headers from our current head till we reach the fcu head, as our current head may not be part of the canonical chain after a reorg.
The main behaviour changes imposed by this are that
A) We must request all headers before we can begin to execute the blocks
B) We are no longer able to keep all headers in memory (in case of long syncs)
Description
BlockSyncState&FullBlockSyncStateObservations & Follow-Up Work
The current implementation downloads all headers first before processing blocks without any sort of checkpoint system to keep progress in case of unexpected failures or shutdowns. When comparing to other implementations this doesn't seem to be much of an issue as header download phase is pretty quick. Having all headers downloaded as a first step can also allow us to download headers non-sequentially (aka divide the chain into batches and fetch them in parallel). Note that this is not as high priority as it would only affect large full syncs, but it may slow down our efforts when using full sync to test vm behaviour. Potential follow up work:
SnapBlockSyncStatestruct can be improved/removedCloses #4717 + Closes #4776