Skip to content

Persist data and metadata #218

@lukemartinlogan

Description

@lukemartinlogan

@AGENTS.md
The Restart procedure needs significant work.

Core::Restart

This override's the container restart method.
This will read all metadata from the persistent metadata log and iterate until the metadata table is reconstructed.
We ignore all metadata in the volatile category of storage, restoring only nonvolatile state (both temporary and long-term).
Volatile state is considered lost.

Core::FlushMetadata

This will store both serialized TagInfo and BlobInfo data structures.
We do not need to store the in-memory maps directly -- only the TagInfo and BlobInfo structures are needed to restore those tables.

Core::FlushData

We do not need to do the target iterations here.
Instead, we should be using AsyncPutBlob with the persistence target parameter stored in the Context object.

Core::PutBlob

We need to update the Context to include persistence target.
After the DPE returns, we will filter out all targets that do not meet the requirement.
We will then proceed as normal during the Put operation.

Metadata

Metadata

Assignees

No one assigned

    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