Skip to content

Implement saving and reading persistent states with parts in S3 #1025

Description

@SmaGMan

Scope

  • Extend the S3 persistent-state loader to support persistent states with parts, not only single-file persistents.
  • First attempt to fetch meta.json from S3; if present, read part_split_depth, derive shard part prefixes, and build the FoundState response with parts. Individual parts should be fetched via separate requests.
  • Extend request arguments in the same way as BlockchainRpcService / BlockchainRpcClient.
  • When saving a persistent state to S3, use PersistentState.CachedState.parts_info to discover parts; each part includes a prefix to load its chunks via read_state_chunk.
  • Save meta.json into S3 so that the persistent state with parts can be discovered and downloaded correctly.

Metadata

Metadata

Assignees

Labels

C-coreCore moduleC-rpcRPC subscriber moduleC-storageStorage moduleP-mediumMedium priorityS-featureSeverity: feature. This is adding a new feature.

Type

No type

Projects

No projects

Relationships

None yet

Development

No branches or pull requests

Issue actions