Skip to content

[Chain Orchestrator] Reconsider chain consolidation pattern #395

New issue
New issue
Open
#396
Open
[Chain Orchestrator] Reconsider chain consolidation pattern#395
#396
Assignees
varun-doshi
@frisitano

Description

@frisitano
frisitano
opened on Oct 27, 2025

Overview

Currently, we consolidate all blocks from the safe block up to the head block in one go. This could result in a large number of blocks being read into memory at a single time, increasing the memory overhead. We should reconsider the pattern we use, potentially doing it in batches of blocks.

rollup-node/crates/chain-orchestrator/src/lib.rs

Lines 1084 to 1107 in f3674d8

} else {
let start_block_number = safe_block_number + 1;
// TODO: Make fetching parallel but ensure concurrency limits are respected.
let mut blocks_to_validate = vec![];
for block_number in start_block_number..=head_block_number {
let block = self
.l2_client
.get_block_by_number(block_number.into())
.full()
.await?
.ok_or(ChainOrchestratorError::L2BlockNotFoundInL2Client(block_number))?
.into_consensus()
.map_transactions(|tx| tx.inner.into_inner());
blocks_to_validate.push(block);
}
self.validate_l1_messages(&blocks_to_validate).await?;
self.database
.update_l1_messages_from_l2_blocks(
blocks_to_validate.into_iter().map(|b| (&b).into()).collect(),
)
.await?;
};

Metadata

Metadata

Assignees

Labels

No labels
No labels

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions