feat: support setting LSM write spec for a table (#3396)

## Summary

Split out from #3354

Adds `LsmWriteSpec` and `Table::set_lsm_write_spec` /
`unset_lsm_write_spec` to
install and clear the spec that selects Lance's MemWAL LSM-style write
path for
`merge_insert`.

`LsmWriteSpec` offers three sharding strategies, all built on Lance's
`InitializeMemWalBuilder`:

- `LsmWriteSpec::bucket(column, num_buckets)` — hash-bucket sharding by
the
  single-column unenforced primary key.
- `LsmWriteSpec::identity(column)` — identity sharding by the raw value
of a
  scalar column.
- `LsmWriteSpec::unsharded()` — a single MemWAL shard.

Each can be refined with `with_maintained_indexes(...)` (indexes the
MemWAL
keeps up to date as rows are appended) and
`with_writer_config_defaults(...)`
(default `ShardWriter` configuration recorded in the MemWAL index, so
every
writer starts from the same defaults). All variants require the table to
have
an unenforced primary key.

- `set_lsm_write_spec` installs the spec by initializing the MemWAL
index;
`unset_lsm_write_spec` removes it (dropping the MemWAL index), reverting
to
  the standard `merge_insert` path. `unset` is idempotent.
- Bindings: Python (`LsmWriteSpec.bucket` / `.identity` / `.unsharded`,
  `set_lsm_write_spec` / `unset_lsm_write_spec`) and TypeScript
  (`setLsmWriteSpec` with `specType` `"bucket"` / `"identity"` /
  `"unsharded"`). `RemoteTable` returns `NotSupported`.

The actual `merge_insert` LSM dispatch and `ShardWriter` write path are
a
follow-up — this PR only installs and clears the spec.
This commit is contained in:
Heng Ge
2026-05-18 00:11:33 -07:00
committed by GitHub
parent 6a431ff0a0
commit 0d30b31998
19 changed files with 1386 additions and 50 deletions

View File

@@ -2397,3 +2397,81 @@ describe("setUnenforcedPrimaryKey", () => {
await expect(table.setUnenforcedPrimaryKey("id")).rejects.toThrow();
});
});
describe("setLsmWriteSpec / unsetLsmWriteSpec", () => {
let tmpDir: tmp.DirResult;
beforeEach(() => {
tmpDir = tmp.dirSync({ unsafeCleanup: true });
});
afterEach(() => tmpDir.removeCallback());
async function makeTable(conn: Connection): Promise<Table> {
return await conn.createEmptyTable(
"t",
new arrow.Schema([new arrow.Field("id", new arrow.Int64(), false)]),
);
}
it("installs and removes a bucket spec", async () => {
const conn = await connect(tmpDir.name);
const table = await makeTable(conn);
await table.setUnenforcedPrimaryKey("id");
await table.setLsmWriteSpec({
specType: "bucket",
column: "id",
numBuckets: 4,
});
await table.unsetLsmWriteSpec();
// A second unset errors — there is no spec left to remove.
await expect(table.unsetLsmWriteSpec()).rejects.toThrow();
// A fresh spec can be installed after unset.
await table.setLsmWriteSpec({
specType: "bucket",
column: "id",
numBuckets: 8,
});
});
it("installs an unsharded spec", async () => {
const conn = await connect(tmpDir.name);
const table = await makeTable(conn);
await table.setUnenforcedPrimaryKey("id");
await table.setLsmWriteSpec({ specType: "unsharded" });
await table.unsetLsmWriteSpec();
});
it("installs an identity spec", async () => {
const conn = await connect(tmpDir.name);
const table = await makeTable(conn);
await table.setUnenforcedPrimaryKey("id");
await table.setLsmWriteSpec({ specType: "identity", column: "id" });
await table.unsetLsmWriteSpec();
});
it("rejects an invalid spec", async () => {
const conn = await connect(tmpDir.name);
const table = await makeTable(conn);
await table.setUnenforcedPrimaryKey("id");
// num_buckets out of range.
await expect(
table.setLsmWriteSpec({
specType: "bucket",
column: "id",
numBuckets: 0,
}),
).rejects.toThrow();
// Column mismatch.
await expect(
table.setLsmWriteSpec({
specType: "bucket",
column: "missing",
numBuckets: 4,
}),
).rejects.toThrow();
});
});

View File

@@ -113,6 +113,7 @@ export {
UpdateOptions,
OptimizeOptions,
Version,
LsmWriteSpec,
ColumnAlteration,
} from "./table";

View File

@@ -106,6 +106,27 @@ export interface Version {
metadata: Record<string, string>;
}
/**
* Specification selecting Lance's MemWAL LSM-style write path for
* `mergeInsert`.
*
* `specType` is `"bucket"`, `"identity"`, or `"unsharded"`. For `"bucket"`,
* `column` and `numBuckets` are required; for `"identity"`, `column` is
* required.
*/
export interface LsmWriteSpec {
/** One of `"bucket"`, `"identity"`, or `"unsharded"`. */
specType: "bucket" | "identity" | "unsharded";
/** Bucket and identity variants: the sharding column. */
column?: string;
/** Bucket variant: the number of buckets, in `[1, 1024]`. */
numBuckets?: number;
/** Names of indexes the MemWAL should keep up to date during writes. */
maintainedIndexes?: string[];
/** Default `ShardWriter` configuration recorded in the MemWAL index. */
writerConfigDefaults?: Record<string, string>;
}
/**
* A Table is a collection of Records in a LanceDB Database.
*
@@ -461,6 +482,42 @@ export abstract class Table {
* @returns {Promise<void>}
*/
abstract setUnenforcedPrimaryKey(columns: string | string[]): Promise<void>;
/**
* Install an {@link LsmWriteSpec} on this table, selecting Lance's MemWAL
* LSM-style write path for future `mergeInsert` calls.
*
* `LsmWriteSpec` chooses one of three sharding strategies via `specType`:
*
* - `"bucket"` — hash-bucket writes by the single-column unenforced primary
* key (`column` and `numBuckets` required).
* - `"identity"` — shard by the raw value of a scalar `column`.
* - `"unsharded"` — route every write to a single shard.
*
* All variants require the table to have an unenforced primary key
* ({@link Table#setUnenforcedPrimaryKey}); bucket sharding additionally
* requires it to be the single column being bucketed.
* @param {LsmWriteSpec} spec The sharding spec to install.
* @returns {Promise<void>}
* @example
* ```ts
* await table.setUnenforcedPrimaryKey("id");
* await table.setLsmWriteSpec({
* specType: "bucket",
* column: "id",
* numBuckets: 16,
* maintainedIndexes: ["id_idx"],
* });
* ```
*/
abstract setLsmWriteSpec(spec: LsmWriteSpec): Promise<void>;
/**
* Remove the {@link LsmWriteSpec} from this table, reverting to the standard
* `mergeInsert` write path.
*
* Errors if no spec is currently set.
* @returns {Promise<void>}
*/
abstract unsetLsmWriteSpec(): Promise<void>;
/** Retrieve the version of the table */
abstract version(): Promise<number>;
@@ -914,6 +971,14 @@ export class LocalTable extends Table {
return await this.inner.setUnenforcedPrimaryKey(cols);
}
async setLsmWriteSpec(spec: LsmWriteSpec): Promise<void> {
return await this.inner.setLsmWriteSpec(spec);
}
async unsetLsmWriteSpec(): Promise<void> {
return await this.inner.unsetLsmWriteSpec();
}
async version(): Promise<number> {
return await this.inner.version();
}

View File

@@ -352,6 +352,23 @@ impl Table {
.default_error()
}
#[napi(catch_unwind)]
pub async fn set_lsm_write_spec(&self, spec: LsmWriteSpec) -> napi::Result<()> {
let native_spec = lancedb::table::LsmWriteSpec::try_from(spec)?;
self.inner_ref()?
.set_lsm_write_spec(native_spec)
.await
.default_error()
}
#[napi(catch_unwind)]
pub async fn unset_lsm_write_spec(&self) -> napi::Result<()> {
self.inner_ref()?
.unset_lsm_write_spec()
.await
.default_error()
}
#[napi(catch_unwind)]
pub async fn version(&self) -> napi::Result<i64> {
self.inner_ref()?
@@ -546,6 +563,63 @@ impl From<lancedb::index::IndexConfig> for IndexConfig {
}
}
/// Specification selecting Lance's MemWAL LSM-style write path for
/// `mergeInsert`.
///
/// `specType` must be `"bucket"`, `"identity"`, or `"unsharded"`. For
/// `"bucket"`, `column` and `numBuckets` are required; for `"identity"`,
/// `column` is required.
#[napi(object)]
#[derive(Clone, Debug)]
pub struct LsmWriteSpec {
/// One of `"bucket"`, `"identity"`, or `"unsharded"`.
pub spec_type: String,
/// Bucket and identity variants: the sharding column.
pub column: Option<String>,
/// Bucket variant: the number of buckets, in `[1, 1024]`.
pub num_buckets: Option<u32>,
/// Names of indexes the MemWAL should keep up to date during writes.
pub maintained_indexes: Option<Vec<String>>,
/// Default `ShardWriter` configuration recorded in the MemWAL index.
pub writer_config_defaults: Option<HashMap<String, String>>,
}
impl TryFrom<LsmWriteSpec> for lancedb::table::LsmWriteSpec {
type Error = napi::Error;
fn try_from(value: LsmWriteSpec) -> napi::Result<Self> {
let maintained = value.maintained_indexes.unwrap_or_default();
let writer_config_defaults = value.writer_config_defaults.unwrap_or_default();
let spec = match value.spec_type.as_str() {
"bucket" => {
let column = value.column.ok_or_else(|| {
napi::Error::from_reason("LsmWriteSpec bucket requires `column`")
})?;
let num_buckets = value.num_buckets.ok_or_else(|| {
napi::Error::from_reason("LsmWriteSpec bucket requires `numBuckets`")
})?;
Self::bucket(column, num_buckets)
}
"identity" => {
let column = value.column.ok_or_else(|| {
napi::Error::from_reason("LsmWriteSpec identity requires `column`")
})?;
Self::identity(column)
}
"unsharded" => Self::unsharded(),
other => {
return Err(napi::Error::from_reason(format!(
"LsmWriteSpec `specType` must be 'bucket', 'identity', or 'unsharded', got '{}'",
other
)));
}
};
Ok(spec
.with_maintained_indexes(maintained)
.with_writer_config_defaults(writer_config_defaults))
}
}
/// Statistics about a compaction operation.
#[napi(object)]
#[derive(Clone, Debug)]