@powersync/service-module-mongodb-storage 0.22.0 → 0.23.0

package/src/storage/implementation/v3/MongoBucketBatchV3.ts

@@ -11,6 +11,7 @@ import { MongoWriteBatch } from '../MongoWriteBatch.js';
 import { stopReplicationStreamPipeline } from '../SyncRuleStateUpdate.js';
 import { PersistedBatch } from '../common/PersistedBatch.js';
 import { SourceRecordStore } from '../common/SourceRecordStore.js';
+import { SyncRuleDocumentBase } from '../models.js';
 import { PersistedBatchV3 } from './PersistedBatchV3.js';
 import { SourceRecordStoreV3 } from './SourceRecordStoreV3.js';
 import { VersionedPowerSyncMongoV3 } from './VersionedPowerSyncMongoV3.js';
@@ -61,7 +62,11 @@ export class MongoBucketBatchV3 extends MongoBucketBatch {
     return this.store;
   }
 
-  protected override onReplicationTransactionFlush(writes: MongoWriteBatch, lastOp: bigint): void {
+  protected override persistedOpHead(stream: SyncRuleDocumentBase): InternalOpId {
+    return (stream as ReplicationStreamDocumentV3).last_persisted_op ?? 0n;
+  }
+
+  protected override onReplicationTransactionFlush(writes: MongoWriteBatch, lastOp: InternalOpId): void {
     // Durably advance the stream-level head of persisted ops within the flush transaction.
     // This ensures a checkpoint created later (even by an empty commit, or by a freshly-appended
     // config that replicates nothing) covers all ops persisted before a potential crash.
@@ -171,7 +176,7 @@ export class MongoBucketBatchV3 extends MongoBucketBatch {
     const session = this.db.client.startSession();
     await using _ = { [Symbol.asyncDispose]: () => session.endSession() };
 
-    await session.withTransaction(async () => {
+    await this.withTransaction(async () => {
       const col = this.db.sourceTables(this.replicationStreamId);
 
       // Find records that overlap by name or relation id.
@@ -241,7 +246,7 @@ export class MongoBucketBatchV3 extends MongoBucketBatch {
           sourceTableFromDocument(doc, context.connectionTag, syncConfig, mapping, eventById)
         )
       };
-    });
+    }, session);
 
     return result!;
   }
@@ -266,156 +271,143 @@ export class MongoBucketBatchV3 extends MongoBucketBatch {
   async commit(lsn: string, options?: storage.BucketBatchCommitOptions): Promise<storage.CheckpointResult> {
     const { createEmptyCheckpoints } = { ...storage.DEFAULT_BUCKET_BATCH_COMMIT_OPTIONS, ...options };
 
-    await this.flush(options);
-
     using _ = this.tracer.span('storage', 'commit');
 
-    const now = new Date();
-
-    await this.db.write_checkpoints.updateMany(
-      {
-        processed_at_lsn: null,
-        'lsns.1': { $lte: lsn }
-      },
-      {
-        $set: {
-          processed_at_lsn: lsn
-        }
-      },
-      {
-        session: this.session
-      }
-    );
-
-    const preUpdateDocument = (await this.db.sync_rules.findOne(
-      {
-        _id: this.replicationStreamId,
-        'sync_configs._id': { $in: this.syncConfigIds }
-      },
-      {
-        session: this.session,
-        projection: {
-          sync_configs: 1,
-          last_persisted_op: 1
-        }
-      }
-    )) as ReplicationStreamDocumentV3 | null;
-
-    const states =
-      preUpdateDocument?.sync_configs?.filter((config) => this.syncConfigIds.some((id) => id.equals(config._id))) ?? [];
-    if (states.length == 0) {
-      throw new ReplicationAssertionError(
-        `Failed to update checkpoint - no matching sync_config for _id: ${this.replicationStreamId}/${this.syncConfigIds
-          .map((id) => id.toHexString())
-          .join(',')}`
+    const checkpoint = async (stream: SyncRuleDocumentBase, lastOp?: InternalOpId) => {
+      const now = new Date();
+      const preUpdateDocument = stream as ReplicationStreamDocumentV3;
+      const writes = this.db.createWriteBatch(this.session, { ordered: false });
+      writes.updateMany(
+        this.db.write_checkpoints,
+        { processed_at_lsn: null, 'lsns.1': { $lte: lsn } },
+        { $set: { processed_at_lsn: lsn } }
       );
-    }
-    // The replication job / MongoBucketBatch must be constructed with all replicating (PROCESSING / ACTIVE)
-    // sync configs in the stream, otherwise we'll get inconsistencies. Configs in other states (e.g. STOP)
-    // remain embedded in the document and are ignored here.
-    const missingSyncConfig = preUpdateDocument!.sync_configs.find(
-      (config) =>
-        [storage.SyncRuleState.PROCESSING, storage.SyncRuleState.ACTIVE].includes(config.state) &&
-        !this.syncConfigIds.some((id) => id.equals(config._id))
-    );
-    if (missingSyncConfig != null) {
-      throw new ReplicationAssertionError(`Replication job not configured for sync config ${missingSyncConfig._id}`);
-    }
 
-    // Effective head of the stream's op sequence.
-    // last_persisted_op is $max-advanced durably in the same transaction as every flush, and this
-    // read uses the same session, so it covers all ops persisted by this batch.
-    const newCheckpoint =
-      preUpdateDocument?.last_persisted_op == null ? 0n : BigInt(preUpdateDocument.last_persisted_op);
-
-    let checkpointBlocked = false;
-    let checkpointCreated = false;
-    let checkpointLogState: unknown = null;
-    const unblockedConfigIds: bson.ObjectId[] = [];
-
-    for (const state of states) {
-      if (state.last_checkpoint != null && state.last_checkpoint > newCheckpoint) {
-        // last_persisted_op is $max-advanced durably in the same transaction as every flush, and
-        // checkpoints are only ever created at that head, so a checkpoint past the head means the
-        // op sequence or the stored state is corrupt.
+      const states =
+        preUpdateDocument?.sync_configs?.filter((config) => this.syncConfigIds.some((id) => id.equals(config._id))) ??
+        [];
+      if (states.length == 0) {
         throw new ReplicationAssertionError(
-          `Invariant violation: sync config ${state._id} has last_checkpoint ${state.last_checkpoint} > stream head ${newCheckpoint}`
+          `Failed to update checkpoint - no matching sync_config for _id: ${this.replicationStreamId}/${this.syncConfigIds
+            .map((id) => id.toHexString())
+            .join(',')}`
         );
       }
+      // The replication job / MongoBucketBatch must be constructed with all replicating (PROCESSING / ACTIVE)
+      // sync configs in the stream, otherwise we'll get inconsistencies. Configs in other states (e.g. STOP)
+      // remain embedded in the document and are ignored here.
+      const missingSyncConfig = preUpdateDocument!.sync_configs.find(
+        (config) =>
+          [storage.SyncRuleState.PROCESSING, storage.SyncRuleState.ACTIVE].includes(config.state) &&
+          !this.syncConfigIds.some((id) => id.equals(config._id))
+      );
+      if (missingSyncConfig != null) {
+        throw new ReplicationAssertionError(`Replication job not configured for sync config ${missingSyncConfig._id}`);
+      }
 
-      const canCheckpoint = canCheckpointState(lsn, {
-        snapshotDone: state.snapshot_done === true,
-        lastCheckpointLsn: state.last_checkpoint_lsn,
-        noCheckpointBefore: state.no_checkpoint_before
-      });
+      // Effective head of the stream's op sequence.
+      // A combined flush supplies its new head; an empty commit uses the fenced persisted head.
+      const newCheckpoint = lastOp ?? this.persistedOpHead(preUpdateDocument);
 
-      if (!canCheckpoint) {
-        checkpointBlocked = true;
-        // Log the first blocked config's state.
-        checkpointLogState ??= {
-          snapshot_done: state.snapshot_done,
-          last_checkpoint_lsn: state.last_checkpoint_lsn,
-          no_checkpoint_before: state.no_checkpoint_before
-        };
-        continue;
-      }
+      let checkpointBlocked = false;
+      let checkpointCreated = false;
+      let checkpointLogState: unknown = null;
+      const unblockedConfigIds: bson.ObjectId[] = [];
 
-      checkpointCreated ||= createEmptyCheckpoints || state.last_checkpoint !== newCheckpoint;
-      unblockedConfigIds.push(state._id);
-    }
+      for (const state of states) {
+        if (state.last_checkpoint != null && state.last_checkpoint > newCheckpoint) {
+          // last_persisted_op is $max-advanced durably in the same transaction as every flush, and
+          // checkpoints are only ever created at that head, so a checkpoint past the head means the
+          // op sequence or the stored state is corrupt.
+          throw new ReplicationAssertionError(
+            `Invariant violation: sync config ${state._id} has last_checkpoint ${state.last_checkpoint} > stream head ${newCheckpoint}`
+          );
+        }
 
-    // Every commit advances the stream's resume position: commit() flushes first, so all
-    // source changes up to this lsn have been persisted, even when checkpoints are blocked.
-    // In the future we could also advance this on flush, when the connector provides the
-    // current position (see setResumeLsn, which connectors may already call after flushing).
-    const resumeLsnUpdate = { resume_lsn: lsn };
-
-    if (unblockedConfigIds.length > 0) {
-      // All unblocked configs get the SAME new value, so we apply it with a single updateOne
-      // (single-document atomicity).
-      const updateSet: Record<string, any> = {
-        last_keepalive_ts: now,
-        last_fatal_error: null,
-        last_fatal_error_ts: null
-      };
-      // Only advance checkpoint fields when an actual (non-empty) checkpoint is created, matching
-      // the previous per-config / v1 behaviour.
-      if (checkpointCreated) {
-        updateSet['sync_configs.$[config].last_checkpoint'] = newCheckpoint;
-        updateSet['sync_configs.$[config].last_checkpoint_lsn'] = lsn;
-        updateSet['last_checkpoint_ts'] = now;
+        const canCheckpoint = canCheckpointState(lsn, {
+          snapshotDone: state.snapshot_done === true,
+          lastCheckpointLsn: state.last_checkpoint_lsn,
+          noCheckpointBefore: state.no_checkpoint_before
+        });
+
+        if (!canCheckpoint) {
+          checkpointBlocked = true;
+          // Log the first blocked config's state.
+          checkpointLogState ??= {
+            snapshot_done: state.snapshot_done,
+            last_checkpoint_lsn: state.last_checkpoint_lsn,
+            no_checkpoint_before: state.no_checkpoint_before
+          };
+          continue;
+        }
+
+        checkpointCreated ||= createEmptyCheckpoints || state.last_checkpoint !== newCheckpoint;
+        unblockedConfigIds.push(state._id);
       }
 
-      await this.db.sync_rules.updateOne(
-        {
-          _id: this.replicationStreamId,
-          'sync_configs._id': { $in: unblockedConfigIds }
-        },
-        { $set: updateSet, $max: resumeLsnUpdate },
-        {
-          session: this.session,
-          arrayFilters: checkpointCreated ? [{ 'config._id': { $in: unblockedConfigIds } }] : undefined
+      // Every commit advances the stream's resume position: commit() flushes first, so all
+      // source changes up to this lsn have been persisted, even when checkpoints are blocked.
+      // In the future we could also advance this on flush, when the connector provides the
+      // current position (see setResumeLsn, which connectors may already call after flushing).
+      const resumeLsnUpdate = {
+        resume_lsn: lsn,
+        ...(lastOp == null ? {} : { last_persisted_op: lastOp })
+      };
+
+      if (unblockedConfigIds.length > 0) {
+        // All unblocked configs get the SAME new value, so we apply it with a single updateOne
+        // (single-document atomicity).
+        const updateSet: Record<string, any> = {
+          last_fatal_error: null,
+          last_fatal_error_ts: null
+        };
+        // Only advance checkpoint fields when an actual (non-empty) checkpoint is created, matching
+        // the previous per-config / v1 behaviour.
+        if (checkpointCreated) {
+          updateSet['sync_configs.$[config].last_checkpoint'] = newCheckpoint;
+          updateSet['sync_configs.$[config].last_checkpoint_lsn'] = lsn;
+          updateSet['last_checkpoint_ts'] = now;
         }
-      );
-    } else {
-      // All selected configs are blocked - only update keepalive/error tracking and the
-      // resume position.
-      await this.db.sync_rules.updateOne(
-        {
-          _id: this.replicationStreamId
-        },
-        {
-          $set: {
-            last_keepalive_ts: now,
-            last_fatal_error: null,
-            last_fatal_error_ts: null
+
+        writes.updateOne(
+          this.db.sync_rules,
+          {
+            _id: this.replicationStreamId,
+            'sync_configs._id': { $in: unblockedConfigIds }
           },
-          $max: resumeLsnUpdate
-        },
-        { session: this.session }
-      );
-    }
+          { $set: updateSet, $max: resumeLsnUpdate },
+          {
+            arrayFilters: checkpointCreated ? [{ 'config._id': { $in: unblockedConfigIds } }] : undefined
+          }
+        );
+      } else {
+        // All selected configs are blocked - only update keepalive/error tracking and the
+        // resume position.
+        writes.updateOne(
+          this.db.sync_rules,
+          {
+            _id: this.replicationStreamId
+          },
+          {
+            $set: {
+              last_fatal_error: null,
+              last_fatal_error_ts: null
+            },
+            $max: resumeLsnUpdate
+          }
+        );
+      }
 
+      await writes.execute();
+      return { checkpointBlocked, checkpointCreated, checkpointLogState, newCheckpoint };
+    };
+    const projection = { sync_configs: 1, last_persisted_op: 1 };
+    const { checkpointBlocked, checkpointCreated, checkpointLogState, newCheckpoint } = await this.flushAndCommit(
+      checkpoint,
+      () => this.withFencedTransaction(() => this.fence(this.session, projection), checkpoint),
+      options,
+      projection
+    );
     if (checkpointBlocked) {
       if (Date.now() - this.lastWaitingLogThrottledV3 > 5_000) {
         this.logger.info(
@@ -446,23 +438,9 @@ export class MongoBucketBatchV3 extends MongoBucketBatch {
 
   async setResumeLsn(lsn: string): Promise<void> {
     using _ = this.tracer.span('storage', 'set_resume_lsn');
-    await this.db.sync_rules.updateOne(
-      {
-        _id: this.replicationStreamId
-      },
-      {
-        $set: {
-          resume_lsn: lsn
-        }
-      },
-      {
-        session: this.session,
-        // Losing occasional resume LSN is fine. That may mean reprocessing
-        // some source changes in some edge cases, which is not an issue since
-        // changes are processed in an idempotent way.
-        writeConcern: { w: 1 }
-      }
-    );
+    // Losing occasional resume LSN would only reprocess source changes.
+    // Keep the lease check atomic, but do not wait for majority replication of this resume hint.
+    await this.updateStreamMetadata({ resume_lsn: { $literal: lsn } }, {}, { w: 1 });
   }
 
   private async autoActivateV3(lsn: string): Promise<void> {
@@ -473,7 +451,7 @@ export class MongoBucketBatchV3 extends MongoBucketBatch {
     const session = this.session;
     let activated = false;
     let needsFutureActivationCheck = true;
-    await session.withTransaction(async () => {
+    await this.withTransaction(async () => {
       // Reset on transaction retries.
       needsFutureActivationCheck = true;
       activated = false;
@@ -561,7 +539,7 @@ export class MongoBucketBatchV3 extends MongoBucketBatch {
       } else if (doc.state == storage.SyncRuleState.ACTIVE && processingStates.length == 0) {
         needsFutureActivationCheck = false;
       }
-    });
+    }, session);
     if (activated) {
       this.logger.info(`Activated new replication stream at ${lsn}`);
       await this.db.notifyCheckpoint();
@@ -571,26 +549,35 @@ export class MongoBucketBatchV3 extends MongoBucketBatch {
     }
   }
 
-  async markAllSnapshotDone(no_checkpoint_before_lsn: string): Promise<void> {
-    await this.db.sync_rules.updateOne(
-      {
-        _id: this.replicationStreamId,
-        'sync_configs._id': { $in: this.syncConfigIds }
-      },
-      {
-        $set: {
-          'sync_configs.$[config].snapshot_done': true,
-          last_keepalive_ts: new Date()
-        },
-        $max: {
-          'sync_configs.$[config].no_checkpoint_before': no_checkpoint_before_lsn
+  private syncConfigMetadataUpdate(syncConfigIds: bson.ObjectId[], set: lib_mongo.mongo.Document) {
+    return {
+      sync_configs: {
+        $map: {
+          input: '$sync_configs',
+          as: 'config',
+          in: {
+            $cond: [
+              { $in: ['$$config._id', { $literal: syncConfigIds }] },
+              { $mergeObjects: ['$$config', set] },
+              '$$config'
+            ]
+          }
         }
-      },
-      {
-        session: this.session,
-        arrayFilters: [{ 'config._id': { $in: this.syncConfigIds } }]
       }
-    );
+    };
+  }
+
+  private snapshotDoneUpdate(no_checkpoint_before_lsn: string) {
+    return this.syncConfigMetadataUpdate(this.syncConfigIds, {
+      snapshot_done: true,
+      no_checkpoint_before: { $max: ['$$config.no_checkpoint_before', { $literal: no_checkpoint_before_lsn }] }
+    });
+  }
+
+  async markAllSnapshotDone(no_checkpoint_before_lsn: string): Promise<void> {
+    await this.updateStreamMetadata(this.snapshotDoneUpdate(no_checkpoint_before_lsn), {
+      'sync_configs._id': { $in: this.syncConfigIds }
+    });
   }
 
   async markSnapshotDone(no_checkpoint_before_lsn: string, options?: { throwOnConflict?: boolean }): Promise<void> {
@@ -614,7 +601,9 @@ export class MongoBucketBatchV3 extends MongoBucketBatch {
         }
       }
 
-      await this.markAllSnapshotDone(no_checkpoint_before_lsn);
+      await this.updateStreamMetadataInTransaction(this.snapshotDoneUpdate(no_checkpoint_before_lsn), {
+        'sync_configs._id': { $in: this.syncConfigIds }
+      });
     });
   }
 
@@ -623,22 +612,9 @@ export class MongoBucketBatchV3 extends MongoBucketBatch {
     if (syncConfigIds.length == 0) {
       return;
     }
-
-    await this.db.sync_rules.updateOne(
-      {
-        _id: this.replicationStreamId,
-        'sync_configs._id': { $in: syncConfigIds }
-      },
-      {
-        $set: {
-          'sync_configs.$[config].snapshot_done': false
-        }
-      },
-      {
-        session: this.session,
-        arrayFilters: [{ 'config._id': { $in: syncConfigIds } }]
-      }
-    );
+    await this.updateStreamMetadata(this.syncConfigMetadataUpdate(syncConfigIds, { snapshot_done: false }), {
+      'sync_configs._id': { $in: syncConfigIds }
+    });
   }
 
   async markTableSnapshotDone(
@@ -672,9 +648,6 @@ export class MongoBucketBatchV3 extends MongoBucketBatch {
             'sync_configs._id': { $in: syncConfigIds }
           },
           {
-            $set: {
-              last_keepalive_ts: new Date()
-            },
             $max: {
               'sync_configs.$[config].no_checkpoint_before': no_checkpoint_before_lsn
             }

package/src/storage/implementation/v3/MongoCompactorV3.ts

@@ -1,18 +1,11 @@
 import { mongo } from '@powersync/lib-service-mongodb';
 import { ReplicationAssertionError, ServiceAssertionError } from '@powersync/lib-services-framework';
-import {
-  acquireSemaphoreAbortable,
-  addChecksums,
-  formatBytes,
-  InternalOpId,
-  storage,
-  utils
-} from '@powersync/service-core';
+import { acquireSemaphoreAbortable, addChecksums, formatBytes, InternalOpId, utils } from '@powersync/service-core';
 import { BucketDefinitionId } from '@powersync/service-sync-rules';
 import { setImmediate } from 'node:timers/promises';
 import { BucketDataDoc } from '../common/BucketDataDoc.js';
 import { BucketDataKey } from '../models.js';
-import { ConcurrentCompactionError, MongoCompactor } from '../MongoCompactor.js';
+import { ConcurrentCompactionError, MongoCompactOptions, MongoCompactor } from '../MongoCompactor.js';
 import { MongoWriteBatch } from '../MongoWriteBatch.js';
 import { cacheKey } from '../OperationBatch.js';
 import { loadBucketDataDocument, maxOpId, serializeBucketData } from './bucket-format.js';
@@ -23,6 +16,7 @@ import {
   bucketStats,
   BucketStatsWithChecksum,
   chooseCompactionKind,
+  chooseRequestedCompactionKind,
   combineAdjacentStats,
   combineChunkStats,
   CompactIntervalConfig,
@@ -33,7 +27,6 @@ import {
   CompactTargetConfig,
   emptyBucketStats,
   firstUncompactedWrite,
-  forcedCompactionKind,
   PendingCompactionGroup,
   readCompactionBatch,
   ScheduledCompactionOptions,
@@ -78,15 +71,17 @@ export class MongoCompactorV3 extends MongoCompactor implements CompactIntervalC
   readonly maxCompactFullIntervalMs: number;
   readonly compactLeaseDurationMs: number;
   readonly maxOpIdCap: InternalOpId | undefined;
+  readonly forceChunkCompaction: boolean;
   private readonly objectStorageUsage: ObjectStorageUsage;
 
-  constructor(bucketStorage: MongoSyncBucketStorageV3, db: VersionedPowerSyncMongoV3, options: storage.CompactOptions) {
+  constructor(bucketStorage: MongoSyncBucketStorageV3, db: VersionedPowerSyncMongoV3, options: MongoCompactOptions) {
     super(bucketStorage, db, options);
     this.minCompactChunkIntervalMs = options.minCompactChunkIntervalMs ?? DEFAULT_MIN_COMPACT_CHUNK_INTERVAL_MS;
     this.minCompactFullIntervalMs = options.minCompactFullIntervalMs ?? DEFAULT_MIN_COMPACT_FULL_INTERVAL_MS;
     this.maxCompactFullIntervalMs = options.maxCompactFullIntervalMs ?? DEFAULT_MAX_COMPACT_FULL_INTERVAL_MS;
     this.compactLeaseDurationMs = options.compactLeaseDurationMs ?? DEFAULT_COMPACT_LEASE_DURATION_MS;
     this.maxOpIdCap = options.maxOpId;
+    this.forceChunkCompaction = options.forceChunkCompaction ?? false;
     this.objectStorageUsage = new ObjectStorageUsage(this.db, this.group_id, createObjectStorageUsageWriterId());
   }
 
@@ -113,7 +108,7 @@ export class MongoCompactorV3 extends MongoCompactor implements CompactIntervalC
       // processes the work that existed when it started.
       await this.compactScheduledBuckets({
         dueAheadMs: DEFAULT_MIN_COMPACT_CHUNK_INTERVAL_MS,
-        forceKind: CompactionKind.Chunks
+        requestedKind: CompactionKind.Chunks
       });
     } else {
       await this.compactScheduledBuckets();
@@ -187,7 +182,7 @@ export class MongoCompactorV3 extends MongoCompactor implements CompactIntervalC
     // the exact initial-replication interval.
     const jobStartedAt = await this.readCompactionTime();
     const dueBefore = new Date(jobStartedAt.getTime() + (options.dueAheadMs ?? 0));
-    const forceKind = options.forceKind;
+    const requestedKind = options.requestedKind;
     const rescheduleNotBefore = new Date(dueBefore.getTime() + 1);
     // Keep accounting documents bounded by workers, not buckets or scan batches.
     const workerUsage = Array.from(
@@ -207,31 +202,31 @@ export class MongoCompactorV3 extends MongoCompactor implements CompactIntervalC
       const scheduled: {
         state: BucketStateDocumentV3;
         decision: CompactionDecision;
-        forcedKind: CompactionKind | null;
+        selectedKind: CompactionKind | null;
       }[] = [];
       for (const state of states) {
         try {
           scheduled.push({
             state,
             decision: chooseCompactionKind(state, batchStartedAt, this),
-            forcedKind: forcedCompactionKind(state, forceKind, this)
+            selectedKind: chooseRequestedCompactionKind(state, requestedKind, this)
           });
         } catch (error) {
           await this.rescheduleFailedBucket(state, rescheduleNotBefore, error);
         }
       }
       const noOpStates = scheduled.filter(
-        ({ state, decision, forcedKind }) =>
-          state.compact_lease == null && (forceKind == null ? decision.kind : forcedKind) == null
+        ({ state, decision, selectedKind }) =>
+          state.compact_lease == null && (requestedKind == null ? decision.kind : selectedKind) == null
       );
       await this.rescheduleUnclaimedBuckets(noOpStates, rescheduleNotBefore);
 
       const processBucket = async (
-        { state, decision, forcedKind }: (typeof scheduled)[number],
+        { state, decision, selectedKind }: (typeof scheduled)[number],
         objectStorageUsage: ObjectStorageUsage,
         chunksOnly = false
       ) => {
-        const kind = forceKind == null ? decision.kind : forcedKind;
+        const kind = requestedKind == null ? decision.kind : selectedKind;
         if (state.compact_lease == null && kind == null) {
           return;
         }
@@ -243,7 +238,9 @@ export class MongoCompactorV3 extends MongoCompactor implements CompactIntervalC
           }
           const claimedDecision = chooseCompactionKind(lease.state, lease.startedAt, this);
           const claimedKind =
-            forceKind == null ? claimedDecision.kind : forcedCompactionKind(lease.state, forceKind, this);
+            requestedKind == null
+              ? claimedDecision.kind
+              : chooseRequestedCompactionKind(lease.state, requestedKind, this);
           if (chunksOnly && claimedKind === CompactionKind.Full) {
             // The decision changed after scanning. Release the lease without
             // rescheduling; the next batch will classify it with a fresh timestamp.
@@ -275,10 +272,10 @@ export class MongoCompactorV3 extends MongoCompactor implements CompactIntervalC
       };
 
       const chunkBuckets = scheduled.filter(
-        ({ decision, forcedKind }) => (forceKind == null ? decision.kind : forcedKind) === CompactionKind.Chunks
+        ({ decision, selectedKind }) => (requestedKind == null ? decision.kind : selectedKind) === CompactionKind.Chunks
       );
       const sequentialBuckets = scheduled.filter(
-        ({ decision, forcedKind }) => (forceKind == null ? decision.kind : forcedKind) !== CompactionKind.Chunks
+        ({ decision, selectedKind }) => (requestedKind == null ? decision.kind : selectedKind) !== CompactionKind.Chunks
       );
       await this.runChunkCompactionWorkers(chunkBuckets, workerUsage, (entry, usage) =>
         processBucket(entry, usage, true)

package/src/storage/implementation/v3/compact-utils.ts

@@ -35,6 +35,7 @@ export interface CompactIntervalConfig {
 }
 export interface CompactTargetConfig {
   readonly maxOpIdCap: InternalOpId | undefined;
+  readonly forceChunkCompaction?: boolean;
 }
 
 export enum CompactionKind {
@@ -51,7 +52,7 @@ export interface ScheduledCompactionOptions {
   /** Process checks scheduled this far after the captured job start. */
   dueAheadMs?: number;
   /** Used by initial replication, which must not run a full compact. */
-  forceKind?: CompactionKind;
+  requestedKind?: CompactionKind;
 }
 
 export class CompactionContext {
@@ -243,8 +244,7 @@ export function chooseCompactionKind(
   // not wake before the exact full-compaction condition is true.
   const fullCheckWithMargin = new Date(fullCheckAt.getTime() + FULL_COMPACT_RESCHEDULE_MARGIN_MS);
   const compacted = state.compacted_state;
-  const chunksSinceCompact = Math.max(0, state.bucket_stats.chunks - (compacted?.chunks ?? 0));
-  const shouldCompactChunks = chunksSinceCompact >= MERGE_CHUNKS_THRESHOLD;
+  const shouldCompactChunks = hasEnoughChunksToMerge(state);
   const canCheckChunks =
     compacted == null || now.getTime() - compacted.at.getTime() >= config.minCompactChunkIntervalMs;
   // Too few new chunks cannot make chunk compaction eligible. Do not poll this
@@ -269,16 +269,25 @@ export function chooseCompactionKind(
   };
 }
 
-export function forcedCompactionKind(
+export function chooseRequestedCompactionKind(
   state: BucketStateDocumentV3,
-  forceKind: CompactionKind | undefined,
+  requestedKind: CompactionKind | undefined,
   config: CompactTargetConfig
 ): CompactionKind | null {
-  if (forceKind == null) {
+  if (requestedKind == null) {
+    return null;
+  }
+  // Initial replication bypasses the interval and never runs full compaction,
+  // but small buckets should still use the normal bulk-rescheduling path.
+  if (requestedKind == CompactionKind.Chunks && !config.forceChunkCompaction && !hasEnoughChunksToMerge(state)) {
     return null;
   }
   const maxOpId = config.maxOpIdCap == null || state.last_op < config.maxOpIdCap ? state.last_op : config.maxOpIdCap;
-  return state.compacted_state == null || state.compacted_state.op_id < maxOpId ? forceKind : null;
+  return state.compacted_state == null || state.compacted_state.op_id < maxOpId ? requestedKind : null;
+}
+
+function hasEnoughChunksToMerge(state: BucketStateDocumentV3): boolean {
+  return state.bucket_stats.chunks - (state.compacted_state?.chunks ?? 0) >= MERGE_CHUNKS_THRESHOLD;
 }
 
 /**

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

@@ -95,7 +95,7 @@ export interface SyncConfigDefinition {
   created_at: Date;
   storage_version: number;
   /**
-   * The related SyncRuleDocumentV3.
+   * The related ReplicationStreamDocumentV3.
    *
    * Note that a specific sync config definition never moves between replication streams. Instead, we can create a new copy for the new replication stream.
    *