feat: add RegionId to FileId

[waynexia]

Signed-off-by: Ruihang Xia [email protected]I hereby agree to the terms of the GreptimeDB CLA.

Refer to a related PR or issue link (optional)

• Cross region file access #6409

What's changed and what's your intention?

Add a RegionId field to FileId struct

PR Checklist

Please convert it to a draft if some of the following conditions are not met.

• I have written the necessary rustdoc comments.
• I have added the necessary unit tests and integration tests.
• This PR requires documentation updates.
• API changes are backward compatible.
• Schema or data changes are backward compatible.

[Copilot]

Pull Request Overview

This PR refactors stale manifest structures and updates file identification by adding a RegionId field to the FileId struct. Key changes include removing the old manifest module, updating API paths and file handling functions to use table_dir instead of region_dir, and replacing FileId::random() calls with FileId::new(region_id).

Reviewed Changes

Copilot reviewed 31 out of 31 changed files in this pull request and generated no comments.

File	Description
src/store-api/src/manifest.rs	Removed stale manifest module
src/store-api/src/lib.rs	Commented out manifest module and added manifest version constants
src/mito2/*	Updated API and tests to use table_dir and new FileId creation methods

Comments suppressed due to low confidence (2)

src/mito2/src/sst/file.rs:48

• The doc comment for FileId declares its string representation as {region_id}_{uuid} but the Display implementation formats it as {region_name}/{uuid}. Please update the documentation or the implementation for consistency.

///

src/mito2/src/sst/file.rs:575

• The test 'test_file_id_serialization' is currently marked as ignored with a TODO comment. It is recommended to either fix or remove the test to ensure adequate test coverage for FileId serialization.

    #[ignore = "TODO: fix this test or remove it"]

[evenyag]

I'm going to refactor this by adding another struct with region id and file id. Adding region id to FileId looks strange.

@v0y4g3r Do you have any suggestions for this new struct? RegionFileId, RegionIdAndFileId, FileIdWithRegion.

[v0y4g3r]

RegionFileId

I'm going to refactor this by adding another struct with region id and file id. Adding region id to FileId looks strange.

@v0y4g3r Do you have any suggestions for this new struct? RegionFileId, RegionIdAndFileId, FileIdWithRegion.

➕1️⃣ for RegionFileId.

[evenyag]

I'm going to refactor this by adding another struct with region id and file id. Adding region id to FileId looks strange.

Done in a3bfb58

I also modified the code that uses both region id and file id to respect the region id in the RegionFileId, so the touched files are more than refactoring. But it should avoid some work to modify related structs later. We can still use derive(Serialize, Deserialize) after this commit.

@v0y4g3r @waynexia

[evenyag]

We have to introduce an enum PathType to compute the region dir in the mito region correctly.

pub enum PathType {
    /// A bare path - the original path of an engine.
    ///
    /// The path prefix is `{table_dir}/{table_id}_{region_sequence}/`.
    Bare,
    /// A path for the data region of a metric engine table.
    ///
    /// The path prefix is `{table_dir}/{table_id}_{region_sequence}/data/`.
    Data,
    /// A path for the metadata region of a metric engine table.
    ///
    /// The path prefix is `{table_dir}/{table_id}_{region_sequence}/metadata/`.
    Metadata,
}

I added the path_type to RegionCreateRequest and RegionOpenRequest. I renamed the region_dir of these requests to table_dir.

Note that when opening a region in the metric engine, we still specify PathType::Bare.

We still need to rename other region_dir fields in the mito crate to table_dir later. This PR is already too large, so I didn't do this refactoring.

[killme2008]

Lots of conflicts @waynexia

[evenyag]

Lots of conflicts @waynexia

I'll address them later. It's ok to review before addressing the conflicts.