Skip to content

fix(server): shard context writers by instance - #231

Merged
beinan merged 1 commit into
mainfrom
codex/issue-204-writer-shards
Aug 9, 2026
Merged

fix(server): shard context writers by instance#231
beinan merged 1 commit into
mainfrom
codex/issue-204-writer-shards

Conversation

@dcfocus

@dcfocus dcfocus commented Aug 4, 2026

Copy link
Copy Markdown
Collaborator

Summary

  • propagate the stable server instance ID into context-store creation and lazy reopening so replicas own distinct MemWAL shards
  • centralize server context options and document that instance identity applies to every server-managed store
  • add a two-replica regression test that verifies cross-shard visibility and distinct physical WAL shards

Context

#216 added writer-owned sharding to the core ContextStore. The server still opened context stores without shard_id, causing every replica to use the shared default shard. This completes the remaining server-side portion of #204 without changing the request API.

Testing

  • cargo test -p lance-context-server (63 passed)
  • cargo fmt --all -- --check
  • cargo clippy -p lance-context-server --all-targets -- -D warnings

Closes #204

@beinan
beinan merged commit 225666b into main Aug 9, 2026
10 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

ContextStore shards MemWAL by data, not by writer — multi-replica writes contend and can hang

2 participants