@powersync/service-module-mongodb-storage 0.15.4 → 0.17.0

package/src/storage/implementation/v3/VersionedPowerSyncMongoV3.ts

@@ -0,0 +1,117 @@
+import { mongo } from '@powersync/lib-service-mongodb';
+import { BucketDefinitionId, ParameterIndexId } from '@powersync/service-sync-rules';
+import { BaseVersionedPowerSyncMongo } from '../common/VersionedPowerSyncMongoBase.js';
+import { CommonSourceTableDocument } from '../models.js';
+import {
+  BucketDataDocumentV3,
+  BucketParameterDocumentV3,
+  BucketStateDocumentV3,
+  CurrentDataDocumentV3,
+  SourceTableDocumentV3,
+  SyncConfigDefinition
+} from './models.js';
+
+export class VersionedPowerSyncMongoV3 extends BaseVersionedPowerSyncMongo {
+  get syncConfigDefinitions() {
+    return this.db.collection<SyncConfigDefinition>('sync_config');
+  }
+
+  sourceRecordsV3(replicationStreamId: number, sourceTableId: mongo.ObjectId): mongo.Collection<CurrentDataDocumentV3> {
+    const collectionName = this.sourceRecordsCollectionName(replicationStreamId, sourceTableId);
+    return this.db.collection<CurrentDataDocumentV3>(collectionName);
+  }
+
+  async listSourceRecordCollectionsV3(replicationStreamId: number): Promise<mongo.Collection<CurrentDataDocumentV3>[]> {
+    return this.listCollectionsByPrefix<CurrentDataDocumentV3>(`source_records_${replicationStreamId}_`);
+  }
+
+  async initializeSourceRecordsCollection(replicationStreamId: number, sourceTableId: mongo.ObjectId) {
+    await this.sourceRecordsV3(replicationStreamId, sourceTableId).createIndex(
+      {
+        pending_delete: 1
+      },
+      {
+        partialFilterExpression: { pending_delete: { $exists: true } },
+        name: 'pending_delete'
+      }
+    );
+  }
+
+  commonSourceTables(replicationStreamId: number): mongo.Collection<CommonSourceTableDocument> {
+    return this.sourceTablesV3(replicationStreamId) as mongo.Collection<CommonSourceTableDocument>;
+  }
+
+  bucketStateV3(replicationStreamId: number): mongo.Collection<BucketStateDocumentV3> {
+    return this.db.collection(`bucket_state_${replicationStreamId}`);
+  }
+
+  parameterIndexV3(
+    replicationStreamId: number,
+    indexId: ParameterIndexId
+  ): mongo.Collection<BucketParameterDocumentV3> {
+    return this.db.collection(`parameter_index_${replicationStreamId}_${indexId}`);
+  }
+
+  sourceTablesV3(replicationStreamId: number): mongo.Collection<SourceTableDocumentV3> {
+    return this.db.collection<SourceTableDocumentV3>(this.sourceTableCollectionName(replicationStreamId));
+  }
+
+  async initializeStreamStorage(replicationStreamId: number) {
+    const sourceTables = this.sourceTablesV3(replicationStreamId);
+    const bucketState = this.bucketStateV3(replicationStreamId);
+    await sourceTables.createIndex(
+      {
+        connection_id: 1,
+        schema_name: 1,
+        table_name: 1,
+        relation_id: 1
+      },
+      {
+        name: 'source_lookup'
+      }
+    );
+    await sourceTables.createIndex(
+      {
+        latest_pending_delete: 1
+      },
+      {
+        partialFilterExpression: { latest_pending_delete: { $exists: true } },
+        name: 'latest_pending_delete'
+      }
+    );
+    await bucketState.createIndex(
+      {
+        last_op: 1
+      },
+      { name: 'bucket_updates', unique: true }
+    );
+    await bucketState.createIndex(
+      {
+        'estimate_since_compact.count': -1
+      },
+      { name: 'dirty_count' }
+    );
+  }
+
+  bucketDataV3(replicationStreamId: number, definitionId: BucketDefinitionId) {
+    return this.db.collection<BucketDataDocumentV3>(`bucket_data_${replicationStreamId}_${definitionId}`);
+  }
+
+  listBucketDataCollectionsV3(replicationStreamId: number) {
+    return this.upstream.listBucketDataCollectionsV3(replicationStreamId);
+  }
+
+  async listParameterIndexCollectionsV3(
+    replicationStreamId: number
+  ): Promise<{ collection: mongo.Collection<BucketParameterDocumentV3>; indexId: ParameterIndexId }[]> {
+    const prefix = `parameter_index_${replicationStreamId}_`;
+    const collections = await this.db.listCollections({ name: new RegExp(`^${prefix}`) }, { nameOnly: true }).toArray();
+
+    return collections
+      .filter((collection) => collection.name.startsWith(prefix))
+      .map((collection) => ({
+        collection: this.db.collection<BucketParameterDocumentV3>(collection.name),
+        indexId: collection.name.slice(prefix.length)
+      }));
+  }
+}

package/src/storage/implementation/v3/models.ts

@@ -0,0 +1,164 @@
+import { InternalOpId, SerializedSyncPlan, SyncRuleState } from '@powersync/service-core';
+import { BucketDefinitionId, ParameterIndexId } from '@powersync/service-sync-rules';
+import * as bson from 'bson';
+import { BucketDataDoc, BucketKey } from '../common/BucketDataDoc.js';
+import {
+  BucketDataDocumentBase,
+  BucketDataKey,
+  BucketParameterDocumentBase,
+  BucketStateDocumentBase,
+  CurrentBucket,
+  ReplicaId,
+  SourceTableDocument,
+  SourceTableKey,
+  SyncRuleCheckpointFields,
+  SyncRuleDocumentBase,
+  TaggedBucketParameterDocument
+} from '../models.js';
+
+/**
+ * Embedded in sync_rules.sync_configs.
+ */
+export interface SyncRuleConfigStateV3 extends SyncRuleCheckpointFields<bigint | null> {
+  _id: bson.ObjectId;
+
+  /**
+   * If false, we cannot create any checkpoints.
+   */
+  snapshot_done: boolean;
+
+  state: SyncRuleState;
+}
+
+/**
+ * Represents the state of a replication stream, in the sync_rules collection.
+ *
+ * Differences from V1:
+ * 1. The static config is moved into a separate sync_configs collection.
+ * 2. The same replication stream may be shared by multiple sync config instances.
+ */
+export interface ReplicationStreamDocumentV3 extends SyncRuleDocumentBase {
+  storage_version: number;
+
+  /**
+   * These contain the checkpoint/state per sync config.
+   *
+   * In common cases we'd have one active config or one active + one processing config,
+   * but the model allows multiple configs in any state.
+   */
+  sync_configs: SyncRuleConfigStateV3[];
+}
+
+/**
+ * Static sync config definition.
+ *
+ * This should be treated as immutable - we don't update this after initial creation.
+ */
+export interface SyncConfigDefinition {
+  _id: bson.ObjectId;
+  created_at: Date;
+  storage_version: number;
+  /**
+   * The related SyncRuleDocumentV3.
+   *
+   * Note that a specific sync config definition never moves between replication streams. Instead, we can create a new copy for the new replication stream.
+   *
+   * When terminating a specific sync config definition, we remove the reference from replication stream -> sync config, but this reference here remains as a historical record.
+   */
+  replication_stream_id: number;
+
+  content: string;
+  serialized_plan?: SerializedSyncPlan | null;
+
+  rule_mapping: {
+    /**
+     * Map of uniqueName -> id, unique per replication stream.
+     */
+    definitions: Record<string, string>;
+    /**
+     * Map of (lookupName, queryId) -> id, unique per replication stream.
+     */
+    parameter_indexes: Record<string, string>;
+  };
+}
+
+export interface CurrentBucketV3 extends CurrentBucket {
+  def: BucketDefinitionId;
+}
+
+export interface RecordedLookupV3 {
+  i: ParameterIndexId;
+  l: bson.Binary;
+}
+
+export interface CurrentDataDocumentV3 {
+  _id: ReplicaId;
+  data: bson.Binary | null;
+  buckets: CurrentBucketV3[];
+  lookups: RecordedLookupV3[];
+  /**
+   * If set, this can be deleted, once there is a consistent checkpoint >= pending_delete.
+   *
+   * This must only be set if buckets = [], lookups = [].
+   */
+  pending_delete?: bigint;
+}
+
+export interface BucketParameterDocumentV3 extends BucketParameterDocumentBase<SourceTableKey> {}
+
+export type BucketDataKeyV3 = BucketDataKey;
+
+export interface BucketDataDocumentV3 extends BucketDataDocumentBase {
+  _id: BucketDataKeyV3;
+}
+
+export function serializeBucketDataV3(document: BucketDataDoc): BucketDataDocumentV3 {
+  const { bucketKey, o } = document;
+  return {
+    _id: {
+      b: bucketKey.bucket,
+      o: o
+    },
+    // List fields directly, so that we don't accidentally persist any unknown fields
+    op: document.op,
+    source_table: document.source_table,
+    source_key: document.source_key,
+    table: document.table,
+    row_id: document.row_id,
+    checksum: document.checksum,
+    data: document.data,
+    target_op: document.target_op
+  };
+}
+
+export function loadBucketDataDocumentV3(
+  context: Pick<BucketKey, 'replicationStreamId' | 'definitionId'>,
+  doc: BucketDataDocumentV3
+): BucketDataDoc {
+  const { _id, ...rest } = doc;
+  return {
+    bucketKey: {
+      ...context,
+      bucket: _id.b
+    },
+    o: _id.o,
+    ...rest
+  };
+}
+
+export function taggedBucketParameterDocumentToV3(document: TaggedBucketParameterDocument): BucketParameterDocumentV3 {
+  const { index: _index, ...rest } = document;
+  return rest as BucketParameterDocumentV3;
+}
+
+export interface SourceTableDocumentV3 extends SourceTableDocument {
+  bucket_data_source_ids: BucketDefinitionId[];
+  parameter_lookup_source_ids: ParameterIndexId[];
+  latest_pending_delete?: InternalOpId | undefined;
+}
+
+export interface BucketStateDocumentV3 extends BucketStateDocumentBase {
+  _id: BucketStateDocumentBase['_id'] & {
+    d: BucketDefinitionId;
+  };
+}

package/src/storage/storage-index.ts

@@ -1,14 +1,17 @@
 export * as test_utils from '../utils/test-utils.js';
 export * from '../utils/util.js';
+export * from './implementation/BucketDefinitionMapping.js';
+export * from './implementation/common/PersistedBatch.js';
+export * from './implementation/createMongoSyncBucketStorage.js';
 export * from './implementation/db.js';
 export * from './implementation/models.js';
-export * from './implementation/MongoBucketBatch.js';
 export * from './implementation/MongoIdSequence.js';
+export * from './implementation/MongoPersistedSyncRules.js';
 export * from './implementation/MongoPersistedSyncRulesContent.js';
 export * from './implementation/MongoStorageProvider.js';
-export * from './implementation/MongoSyncBucketStorage.js';
 export * from './implementation/MongoSyncRulesLock.js';
 export * from './implementation/OperationBatch.js';
-export * from './implementation/PersistedBatch.js';
+export * from './implementation/v1/models.js';
+export * from './implementation/v3/models.js';
 export * from './MongoBucketStorage.js';
 export * from './MongoReportStorage.js';

package/src/utils/util.ts

@@ -1,10 +1,12 @@
+import * as lib_mongo from '@powersync/lib-service-mongodb';
 import { mongo } from '@powersync/lib-service-mongodb';
-import { ServiceAssertionError } from '@powersync/lib-services-framework';
+import { ReplicationAbortedError, ServiceAssertionError } from '@powersync/lib-services-framework';
 import { storage, utils } from '@powersync/service-core';
 import * as bson from 'bson';
 import * as crypto from 'crypto';
+import * as timers from 'node:timers/promises';
 import * as uuid from 'uuid';
-import { BucketDataDocument } from '../storage/implementation/models.js';
+import { BucketDataDoc } from '../storage/implementation/common/BucketDataDoc.js';
 
 export function idPrefixFilter<T>(prefix: Partial<T>, rest: (keyof T)[]): mongo.Condition<T> {
   let filter = {
@@ -69,10 +71,10 @@ export async function readSingleBatch<T>(cursor: mongo.AbstractCursor<T>): Promi
   }
 }
 
-export function mapOpEntry(row: BucketDataDocument): utils.OplogEntry {
+export function mapOpEntry(row: BucketDataDoc): utils.OplogEntry {
   if (row.op == 'PUT' || row.op == 'REMOVE') {
     return {
-      op_id: utils.internalToExternalOpId(row._id.o),
+      op_id: utils.internalToExternalOpId(row.o),
       op: row.op,
       object_type: row.table,
       object_id: row.row_id,
@@ -84,7 +86,7 @@ export function mapOpEntry(row: BucketDataDocument): utils.OplogEntry {
     // MOVE, CLEAR
 
     return {
-      op_id: utils.internalToExternalOpId(row._id.o),
+      op_id: utils.internalToExternalOpId(row.o),
       op: row.op,
       checksum: Number(row.checksum)
     };
@@ -129,6 +131,33 @@ export function setSessionSnapshotTime(session: mongo.ClientSession, time: bson.
   }
 }
 
+export async function retryOnMongoMaxTimeMSExpired<T>(
+  operation: () => Promise<T>,
+  options: {
+    signal?: AbortSignal;
+    abortMessage?: string;
+    retryDelayMs: number;
+    onRetry?: (retryCount: number) => void;
+  }
+): Promise<T> {
+  let retryCount = 0;
+  while (true) {
+    if (options.signal?.aborted) {
+      throw new ReplicationAbortedError(options.abortMessage ?? 'Aborted MongoDB operation', options.signal.reason);
+    }
+    try {
+      return await operation();
+    } catch (e) {
+      if (!lib_mongo.isMongoServerError(e) || e.codeName !== 'MaxTimeMSExpired') {
+        throw e;
+      }
+      retryCount += 1;
+      options.onRetry?.(retryCount);
+      await timers.setTimeout(options.retryDelayMs);
+    }
+  }
+}
+
 export const createPaginatedConnectionQuery = async <T extends mongo.Document>(
   query: mongo.Filter<T>,
   collection: mongo.Collection<T>,

package/test/src/storage_compacting.test.ts

@@ -1,7 +1,7 @@
+import { VersionedPowerSyncMongoV3 } from '@module/storage/implementation/v3/VersionedPowerSyncMongoV3.js';
 import { storage, SyncRulesBucketStorage, updateSyncRulesFromYaml } from '@powersync/service-core';
 import { bucketRequest, register, test_utils } from '@powersync/service-core-tests';
 import { describe, expect, test } from 'vitest';
-import { MongoCompactor } from '../../src/storage/implementation/MongoCompactor.js';
 import { INITIALIZED_MONGO_STORAGE_FACTORY } from './util.js';
 
 describe('Mongo Sync Bucket Storage Compact', () => {
@@ -64,9 +64,16 @@ bucket_definitions:
 
     test('full compact', async () => {
       const { bucketStorage, checkpoint, factory, syncRules } = await setup();
+      const storageDb = bucketStorage.db;
 
       // Simulate bucket_state from old version not being available
-      await factory.db.bucket_state.deleteMany({});
+      if (storageDb.storageConfig.incrementalReprocessing) {
+        // This should actually never happen on V3, but we test this anyway.
+        // Can remove this if it causes issues in the future.
+        await (storageDb as VersionedPowerSyncMongoV3).bucketStateV3(bucketStorage.group_id).deleteMany({});
+      } else {
+        await factory.db.bucket_state.deleteMany({});
+      }
 
       await bucketStorage.compact({
         clearBatchLimit: 200,
@@ -95,10 +102,10 @@ bucket_definitions:
     });
 
     test('populatePersistentChecksumCache', async () => {
-      // Populate old sync rules version
+      // Populate old replication stream
       const { factory } = await setup();
 
-      // Now populate another version (bucket definition name changed)
+      // Now populate another replication stream (bucket definition name changed)
       const syncRules = await factory.updateSyncRules(
         updateSyncRulesFromYaml(`
 bucket_definitions:
@@ -108,6 +115,7 @@ bucket_definitions:
     `)
       );
       const bucketStorage = factory.getInstance(syncRules);
+      const storageDb = (bucketStorage as any).db;
 
       await populate(bucketStorage, 2);
       const { checkpoint } = await bucketStorage.getCheckpoint();
@@ -158,35 +166,54 @@ bucket_definitions:
     `)
       );
       const bucketStorage = factory.getInstance(syncRules);
+      const storageDb = bucketStorage.db;
 
       // This simulates bucket_state created using bigint bytes.
       // This typically happens when buckets get very large (> 2GiB). We don't want to create that much
       // data in the tests, so we directly insert the bucket_state here.
-      await factory.db.bucket_state.insertOne({
-        _id: {
-          g: bucketStorage.group_id,
-          b: 'global[]'
-        },
-        last_op: 5n,
-        compacted_state: {
-          op_id: 3n,
-          count: 3,
-          checksum: 0n,
-          bytes: 7n
-        },
-        estimate_since_compact: {
-          count: 2,
-          bytes: 5n
-        }
-      });
-
-      // This test uses a couple of internal APIs of the compactor - there is no simple way
-      // to test this using the current public APIs.
-      const compactor = new MongoCompactor(bucketStorage, (bucketStorage as any).db, {
-        maxOpId: 5n
-      });
-
-      const dirtyBuckets = (compactor as any).dirtyBucketBatches({
+      if (storageDb.storageConfig.incrementalReprocessing) {
+        const bucketStateCollection = (storageDb as VersionedPowerSyncMongoV3).bucketStateV3(bucketStorage.group_id);
+        await bucketStateCollection.insertOne({
+          _id: {
+            d: '1',
+            b: 'global[]'
+          },
+          last_op: 5n,
+          compacted_state: {
+            op_id: 3n,
+            count: 3,
+            checksum: 0n,
+            bytes: 7n
+          },
+          estimate_since_compact: {
+            count: 2,
+            bytes: 5n
+          }
+        });
+      } else {
+        await factory.db.bucket_state.insertOne({
+          _id: {
+            g: bucketStorage.group_id,
+            b: 'global[]'
+          },
+          last_op: 5n,
+          compacted_state: {
+            op_id: 3n,
+            count: 3,
+            checksum: 0n,
+            bytes: 7n
+          },
+          estimate_since_compact: {
+            count: 2,
+            bytes: 5n
+          }
+        });
+      }
+
+      // This test uses a couple of "internal" APIs of the compactor.
+      const compactor = bucketStorage.createMongoCompactor({ maxOpId: 5n });
+
+      const dirtyBuckets = compactor.dirtyBucketBatches({
         minBucketChanges: 1,
         minChangeRatio: 0.39
       });
@@ -205,6 +232,7 @@ bucket_definitions:
       expect(checksumBuckets).toEqual([
         {
           bucket: 'global[]',
+          definitionId: storageDb.storageConfig.incrementalReprocessing ? '1' : null,
           estimatedCount: 5
         }
       ]);