@powersync/service-module-mongodb-storage 0.17.0 → 0.18.1

package/src/storage/implementation/v1/MongoBucketBatchV1.ts

@@ -1,5 +1,5 @@
 import { ReplicationAssertionError } from '@powersync/lib-services-framework';
-import { ColumnDescriptor, SourceTable, storage } from '@powersync/service-core';
+import { ColumnDescriptor, InternalOpId, SourceTable, storage } from '@powersync/service-core';
 import * as bson from 'bson';
 import { mongoTableId } from '../../../utils/util.js';
 import { calculateCheckpointState } from '../CheckpointState.js';
@@ -9,22 +9,35 @@ import { SourceRecordStore } from '../common/SourceRecordStore.js';
 import { PersistedBatchV1 } from './PersistedBatchV1.js';
 import { SourceRecordStoreV1 } from './SourceRecordStoreV1.js';
 import { VersionedPowerSyncMongoV1 } from './VersionedPowerSyncMongoV1.js';
-import { SourceTableDocumentV1, SyncRuleDocumentV1 } from './models.js';
+import { SyncRuleDocumentV1 } from './models.js';
 
 export class MongoBucketBatchV1 extends MongoBucketBatch {
-  declare public readonly db: VersionedPowerSyncMongoV1;
+  declare readonly db: VersionedPowerSyncMongoV1;
 
   private readonly store: SourceRecordStore;
   private needsActivation = true;
   private lastWaitingLogThrottled = 0;
 
+  /**
+   * The last persisted op that is not yet covered by a checkpoint, if any.
+   *
+   * Seeded from the persisted keepalive_op, so that ops persisted before a restart can still be
+   * included in the next checkpoint.
+   */
+  private persisted_op: InternalOpId | null = null;
+
   constructor(options: MongoBucketBatchOptions) {
     super(options);
-    this.store = new SourceRecordStoreV1(this.db, this.group_id);
+    this.persisted_op = options.keepaliveOp ?? null;
+    this.store = new SourceRecordStoreV1(this.db, this.replicationStreamId);
+  }
+
+  protected override recordPersistedOp(lastOp: InternalOpId): void {
+    this.persisted_op = lastOp;
   }
 
   protected createPersistedBatch(writtenSize: number): PersistedBatch {
-    return new PersistedBatchV1(this.db, this.group_id, this.mapping, writtenSize, {
+    return new PersistedBatchV1(this.db, this.replicationStreamId, this.mapping, writtenSize, {
       logger: this.logger
     });
   }
@@ -32,13 +45,12 @@ export class MongoBucketBatchV1 extends MongoBucketBatch {
   protected get sourceRecordStore(): SourceRecordStore {
     return this.store;
   }
-
   protected async cleanupDroppedSourceTables(_tables: SourceTable[]) {
     // No-op for V1: source records live in a shared collection.
   }
 
   async resolveTables(options: storage.ResolveTablesOptions): Promise<storage.ResolveTablesResult> {
-    const syncRules = options.syncRules ?? this.sync_rules;
+    const syncRules = options.parsedSyncConfig?.hydratedSyncConfig ?? this.sync_rules;
     const { connection_id, source } = options;
     const { schema, name, objectId, replicaIdColumns, connectionTag, sendsCompleteRows } = source;
 
@@ -50,9 +62,9 @@ export class MongoBucketBatchV1 extends MongoBucketBatch {
 
     let result: storage.ResolveTablesResult | null = null;
     await this.db.client.withSession(async (session) => {
-      const col = this.db.commonSourceTables(this.group_id);
+      const col = this.db.sourceTablesV1(this.replicationStreamId);
       const filter: any = {
-        group_id: this.group_id,
+        group_id: this.replicationStreamId,
         connection_id,
         schema_name: schema,
         table_name: name,
@@ -66,7 +78,7 @@ export class MongoBucketBatchV1 extends MongoBucketBatch {
       if (doc == null) {
         doc = {
           _id: options.idGenerator ? (options.idGenerator() as bson.ObjectId) : new bson.ObjectId(),
-          group_id: this.group_id,
+          group_id: this.replicationStreamId,
           connection_id,
           relation_id: objectId,
           schema_name: schema,
@@ -107,7 +119,7 @@ export class MongoBucketBatchV1 extends MongoBucketBatch {
       const truncate = await col
         .find(
           {
-            group_id: this.group_id,
+            group_id: this.replicationStreamId,
             connection_id,
             _id: { $ne: doc._id },
             $or: truncateFilter
@@ -145,13 +157,13 @@ export class MongoBucketBatchV1 extends MongoBucketBatch {
   }
 
   async getSourceTableStatus(table: storage.SourceTable): Promise<storage.SourceTable | null> {
-    const doc = (await this.db.commonSourceTables(this.group_id).findOne(
+    const doc = await this.db.sourceTablesV1(this.replicationStreamId).findOne(
       {
-        group_id: this.group_id,
+        group_id: this.replicationStreamId,
         _id: mongoTableId(table.id)
       },
       { session: this.session }
-    )) as SourceTableDocumentV1 | null;
+    );
     if (doc == null) {
       return null;
     }
@@ -243,7 +255,7 @@ export class MongoBucketBatchV1 extends MongoBucketBatch {
     };
 
     const preUpdateDocument = (await this.db.sync_rules.findOneAndUpdate(
-      { _id: this.group_id },
+      { _id: this.replicationStreamId },
       [
         {
           $set: {
@@ -296,7 +308,7 @@ export class MongoBucketBatchV1 extends MongoBucketBatch {
 
     if (preUpdateDocument == null) {
       throw new ReplicationAssertionError(
-        'Failed to update checkpoint - no matching sync_rules document for _id: ' + this.group_id
+        'Failed to update checkpoint - no matching sync_rules document for _id: ' + this.replicationStreamId
       );
     }
 
@@ -330,7 +342,6 @@ export class MongoBucketBatchV1 extends MongoBucketBatch {
       await this.autoActivate(lsn);
       await this.db.notifyCheckpoint();
       this.persisted_op = null;
-      this.last_checkpoint_lsn = lsn;
       if (checkpointState.newLastCheckpoint != null) {
         await this.sourceRecordStore.postCommitCleanup(checkpointState.newLastCheckpoint, this.logger);
       }
@@ -349,7 +360,7 @@ export class MongoBucketBatchV1 extends MongoBucketBatch {
   async setResumeLsn(lsn: string): Promise<void> {
     await this.db.sync_rules.updateOne(
       {
-        _id: this.group_id
+        _id: this.replicationStreamId
       },
       {
         $set: {
@@ -363,7 +374,7 @@ export class MongoBucketBatchV1 extends MongoBucketBatch {
   async markAllSnapshotDone(no_checkpoint_before_lsn: string): Promise<void> {
     await this.db.sync_rules.updateOne(
       {
-        _id: this.group_id
+        _id: this.replicationStreamId
       },
       {
         $set: {
@@ -381,17 +392,24 @@ export class MongoBucketBatchV1 extends MongoBucketBatch {
   async markSnapshotDone(no_checkpoint_before_lsn: string, options?: { throwOnConflict?: boolean }): Promise<void> {
     await this.withTransaction(async () => {
       // Protect against race conditions
-      const count = await this.db.commonSourceTables(this.group_id).countDocuments(
-        {
-          group_id: this.group_id,
-          snapshot_done: false
-        },
-        { session: this.session }
-      );
-      if (count > 0) {
+      const blockingTables = await this.db
+        .sourceTablesV1(this.replicationStreamId)
+        .find(
+          {
+            group_id: this.replicationStreamId,
+            snapshot_done: false
+          },
+          {
+            session: this.session,
+            projection: { schema_name: 1, table_name: 1 }
+          }
+        )
+        .toArray();
+
+      if (blockingTables.length > 0) {
         if (options?.throwOnConflict ?? true) {
           throw new ReplicationAssertionError(
-            `Cannot mark snapshot done while ${count} source table${count == 1 ? '' : 's'} still require snapshotting`
+            `Cannot mark snapshot done while source tables still require snapshotting. Tables: ${blockingTables.map((t) => `${t.schema_name}.${t.table_name}`).join(', ')}`
           );
         } else {
           return;
@@ -405,7 +423,7 @@ export class MongoBucketBatchV1 extends MongoBucketBatch {
   async markTableSnapshotRequired(_table: storage.SourceTable): Promise<void> {
     await this.db.sync_rules.updateOne(
       {
-        _id: this.group_id
+        _id: this.replicationStreamId
       },
       {
         $set: {
@@ -424,7 +442,7 @@ export class MongoBucketBatchV1 extends MongoBucketBatch {
     const ids = tables.map((table) => mongoTableId(table.id));
 
     await this.withTransaction(async () => {
-      await this.db.commonSourceTables(this.group_id).updateMany(
+      await this.db.sourceTablesV1(this.replicationStreamId).updateMany(
         { _id: { $in: ids } },
         {
           $set: {
@@ -440,7 +458,7 @@ export class MongoBucketBatchV1 extends MongoBucketBatch {
       if (no_checkpoint_before_lsn != null) {
         await this.db.sync_rules.updateOne(
           {
-            _id: this.group_id
+            _id: this.replicationStreamId
           },
           {
             $set: {
@@ -469,11 +487,14 @@ export class MongoBucketBatchV1 extends MongoBucketBatch {
     const session = this.session;
     let activated = false;
     await session.withTransaction(async () => {
-      const doc = (await this.db.sync_rules.findOne({ _id: this.group_id }, { session })) as SyncRuleDocumentV1;
+      const doc = (await this.db.sync_rules.findOne(
+        { _id: this.replicationStreamId },
+        { session }
+      )) as SyncRuleDocumentV1;
       if (doc && doc.state == storage.SyncRuleState.PROCESSING && doc.snapshot_done && doc.last_checkpoint != null) {
         await this.db.sync_rules.updateOne(
           {
-            _id: this.group_id
+            _id: this.replicationStreamId
           },
           {
             $set: {
@@ -485,7 +506,7 @@ export class MongoBucketBatchV1 extends MongoBucketBatch {
 
         await this.db.sync_rules.updateMany(
           {
-            _id: { $ne: this.group_id },
+            _id: { $ne: this.replicationStreamId },
             state: { $in: [storage.SyncRuleState.ACTIVE, storage.SyncRuleState.ERRORED] }
           },
           {

package/src/storage/implementation/v1/MongoChecksumsV1.ts

@@ -1,11 +1,20 @@
+import * as lib_mongo from '@powersync/lib-service-mongodb';
 import {
+  addPartialChecksums,
   bson,
   BucketChecksum,
   FetchPartialBucketChecksum,
   InternalOpId,
-  PartialChecksumMap
+  isPartialChecksum,
+  PartialChecksumMap,
+  PartialOrFullChecksum
 } from '@powersync/service-core';
-import { FetchPartialBucketChecksumByBucket, MongoChecksums } from '../MongoChecksums.js';
+import {
+  checksumFromAggregate,
+  DEFAULT_OPERATION_BATCH_LIMIT,
+  FetchPartialBucketChecksumByBucket,
+  MongoChecksums
+} from '../MongoChecksums.js';
 import { VersionedPowerSyncMongoV1 } from './VersionedPowerSyncMongoV1.js';
 
 export class MongoChecksumsV1 extends MongoChecksums {
@@ -14,7 +23,8 @@ export class MongoChecksumsV1 extends MongoChecksums {
   async computePartialChecksumsDirectByBucket(
     batch: FetchPartialBucketChecksumByBucket[]
   ): Promise<PartialChecksumMap> {
-    return this.computePartialChecksumsForCollection(batch, this.db.bucketDataV1, (request) => ({
+    const collection = this.db.bucketDataV1;
+    const createFilter = (request: FetchPartialBucketChecksumByBucket) => ({
       _id: {
         $gt: {
           g: this.group_id,
@@ -27,7 +37,130 @@ export class MongoChecksumsV1 extends MongoChecksums {
           o: request.end
         }
       }
-    }));
+    });
+
+    const batchLimit = this.options?.operationBatchLimit ?? DEFAULT_OPERATION_BATCH_LIMIT;
+
+    // Map requests by bucket. We adjust this as we get partial results.
+    let requests = new Map<string, FetchPartialBucketChecksumByBucket>();
+    for (let request of batch) {
+      requests.set(request.bucket, request);
+    }
+
+    const partialChecksums = new Map<string, PartialOrFullChecksum>();
+
+    while (requests.size > 0) {
+      const filters = Array.from(requests.values(), createFilter);
+
+      // Historically, checksum may be stored as 'int' or 'double'.
+      // More recently, this should be a 'long'.
+      // $toLong ensures that we always sum it as a long, avoiding inaccuracies in the calculations.
+      const checksumLong = this.storageConfig.longChecksums ? '$checksum' : { $toLong: '$checksum' };
+
+      // Aggregate over a max of `batchLimit` operations at a time.
+      // Let's say we have 3 buckets (A, B, C), each with 10 operations, and our batch limit is 12.
+      // Then we'll do three batches:
+      // 1. Query: A[1-end], B[1-end], C[1-end]
+      //    Returns: A[1-10], B[1-2]
+      // 2. Query: B[3-end], C[1-end]
+      //    Returns: B[3-10], C[1-4]
+      // 3. Query: C[5-end]
+      //    Returns: C[5-10]
+      const aggregate = await collection
+        .aggregate(
+          [
+            {
+              $match: {
+                $or: filters
+              }
+            },
+            // sort and limit _before_ grouping
+            { $sort: { _id: 1 } },
+            { $limit: batchLimit },
+            {
+              $group: {
+                _id: '$_id.b',
+                checksum_total: { $sum: checksumLong },
+                count: { $sum: 1 },
+                has_clear_op: {
+                  $max: {
+                    $cond: [{ $eq: ['$op', 'CLEAR'] }, 1, 0]
+                  }
+                },
+                last_op: { $max: '$_id.o' }
+              }
+            },
+            // Sort the aggregated results (100 max, so should be fast).
+            // This is important to identify which buckets we have partial data for.
+            { $sort: { _id: 1 } }
+          ],
+          {
+            session: undefined,
+            readConcern: 'snapshot',
+            maxTimeMS: lib_mongo.MONGO_CHECKSUM_TIMEOUT_MS
+          }
+        )
+        .toArray()
+        .catch((e) => {
+          throw lib_mongo.mapQueryError(e, 'while reading checksums');
+        });
+
+      let batchCount = 0;
+      let limitReached = false;
+      for (let doc of aggregate) {
+        const bucket = doc._id;
+        const checksum = checksumFromAggregate(doc);
+
+        const existing = partialChecksums.get(bucket);
+        if (existing != null) {
+          partialChecksums.set(bucket, addPartialChecksums(bucket, existing, checksum));
+        } else {
+          partialChecksums.set(bucket, checksum);
+        }
+
+        batchCount += doc.count;
+        if (batchCount == batchLimit) {
+          // Limit reached. Request more in the next batch.
+          // Note that this only affects the _last_ bucket in a batch.
+          limitReached = true;
+          const req = requests.get(bucket);
+          requests.set(bucket, {
+            ...req!,
+            start: doc.last_op
+          });
+        } else {
+          // All done for this bucket
+          requests.delete(bucket);
+        }
+      }
+      if (!limitReached) {
+        break;
+      }
+    }
+
+    return new Map<string, PartialOrFullChecksum>(
+      batch.map((request) => {
+        const bucket = request.bucket;
+        let partialChecksum = partialChecksums.get(bucket);
+        // Could be null if we got no data
+        if (partialChecksum == null) {
+          partialChecksum = {
+            bucket,
+            partialCount: 0,
+            partialChecksum: 0
+          };
+        }
+        if (request.start == null && isPartialChecksum(partialChecksum)) {
+          partialChecksum = {
+            bucket,
+            count: partialChecksum.partialCount,
+            checksum: partialChecksum.partialChecksum
+          };
+        }
+
+        return [bucket, partialChecksum];
+      })
+    );
   }
 
   protected async fetchPreStates(