@powersync/service-core 0.4.1 → 0.5.0

package/src/entry/entry-index.ts

@@ -3,3 +3,4 @@ export * from './commands/config-command.js';
 export * from './commands/migrate-action.js';
 export * from './commands/start-action.js';
 export * from './commands/teardown-action.js';
+export * from './commands/compact-action.js';

package/src/routes/endpoints/socket-route.ts

@@ -7,6 +7,7 @@ import { streamResponse } from '../../sync/sync.js';
 import * as util from '../../util/util-index.js';
 import { SocketRouteGenerator } from '../router-socket.js';
 import { SyncRoutes } from './sync-stream.js';
+import { RequestTracker } from '../../sync/RequestTracker.js';
 
 export const syncStreamReactive: SocketRouteGenerator = (router) =>
   router.reactiveStream<util.StreamingSyncRequest, any>(SyncRoutes.STREAM, {
@@ -66,6 +67,7 @@ export const syncStreamReactive: SocketRouteGenerator = (router) =>
       });
 
       Metrics.getInstance().concurrent_connections.add(1);
+      const tracker = new RequestTracker();
       try {
         for await (const data of streamResponse({
           storage,
@@ -79,6 +81,7 @@ export const syncStreamReactive: SocketRouteGenerator = (router) =>
             // RSocket handles keepalive events by default
             keep_alive: false
           },
+          tracker,
           signal: controller.signal
         })) {
           if (data == null) {
@@ -94,7 +97,7 @@ export const syncStreamReactive: SocketRouteGenerator = (router) =>
             const serialized = serialize(data) as Buffer;
             responder.onNext({ data: serialized }, false);
             requestedN--;
-            Metrics.getInstance().data_synced_bytes.add(serialized.length);
+            tracker.addDataSynced(serialized.length);
           }
 
           if (requestedN <= 0) {
@@ -126,6 +129,11 @@ export const syncStreamReactive: SocketRouteGenerator = (router) =>
         responder.onComplete();
         removeStopHandler();
         disposer();
+        logger.info(`Sync stream complete`, {
+          user_id: syncParams.user_id,
+          operations_synced: tracker.operationsSynced,
+          data_synced_bytes: tracker.dataSyncedBytes
+        });
         Metrics.getInstance().concurrent_connections.add(-1);
       }
     }

package/src/routes/endpoints/sync-stream.ts

@@ -8,6 +8,7 @@ import * as util from '../../util/util-index.js';
 import { Metrics } from '../../metrics/Metrics.js';
 import { authUser } from '../auth.js';
 import { routeDefinition } from '../router.js';
+import { RequestTracker } from '../../sync/RequestTracker.js';
 
 export enum SyncRoutes {
   STREAM = '/sync/stream'
@@ -43,6 +44,7 @@ export const syncStreamed = routeDefinition({
       });
     }
     const controller = new AbortController();
+    const tracker = new RequestTracker();
     try {
       Metrics.getInstance().concurrent_connections.add(1);
       const stream = Readable.from(
@@ -53,9 +55,11 @@ export const syncStreamed = routeDefinition({
               params,
               syncParams,
               token: payload.context.token_payload!,
+              tracker,
               signal: controller.signal
             })
-          )
+          ),
+          tracker
         ),
         { objectMode: false, highWaterMark: 16 * 1024 }
       );
@@ -86,6 +90,11 @@ export const syncStreamed = routeDefinition({
         afterSend: async () => {
           controller.abort();
           Metrics.getInstance().concurrent_connections.add(-1);
+          logger.info(`Sync stream complete`, {
+            user_id: syncParams.user_id,
+            operations_synced: tracker.operationsSynced,
+            data_synced_bytes: tracker.dataSyncedBytes
+          });
         }
       });
     } catch (ex) {

package/src/storage/BucketStorage.ts

@@ -228,7 +228,7 @@ export interface SyncRulesBucketStorage {
     checkpoint: util.OpId,
     dataBuckets: Map<string, string>,
     options?: BucketDataBatchOptions
-  ): AsyncIterable<util.SyncBucketData>;
+  ): AsyncIterable<SyncBucketDataBatch>;
 
   /**
    * Compute checksums for a given list of buckets.
@@ -266,6 +266,8 @@ export interface SyncRulesBucketStorage {
    * Errors are cleared on commit.
    */
   reportError(e: any): Promise<void>;
+
+  compact(options?: CompactOptions): Promise<void>;
 }
 
 export interface SyncRuleStatus {
@@ -388,6 +390,11 @@ export interface SaveDelete {
   after?: undefined;
 }
 
+export interface SyncBucketDataBatch {
+  batch: util.SyncBucketData;
+  targetOp: bigint | null;
+}
+
 export function mergeToast(record: ToastableSqliteRow, persisted: ToastableSqliteRow): ToastableSqliteRow {
   const newRecord: ToastableSqliteRow = {};
   for (let key in record) {
@@ -399,3 +406,24 @@ export function mergeToast(record: ToastableSqliteRow, persisted: ToastableSqlit
   }
   return newRecord;
 }
+
+export interface CompactOptions {
+  /**
+   * Heap memory limit for the compact process.
+   *
+   * Add around 64MB to this to determine the "--max-old-space-size" argument.
+   * Add another 80MB to get RSS usage / memory limits.
+   */
+  memoryLimitMB?: number;
+
+  /**
+   * If specified, ignore any operations newer than this when compacting.
+   *
+   * This is primarily for tests, where we want to test compacting at a specific
+   * point.
+   *
+   * This can also be used to create a "safe buffer" of recent operations that should
+   * not be compacted, to avoid invalidating checkpoints in use.
+   */
+  maxOpId?: bigint;
+}

package/src/storage/mongo/MongoCompactor.ts

@@ -0,0 +1,356 @@
+import { logger } from '@powersync/lib-services-framework';
+import { AnyBulkWriteOperation, MaxKey, MinKey } from 'mongodb';
+import { addChecksums } from '../../util/utils.js';
+import { PowerSyncMongo } from './db.js';
+import { BucketDataDocument, BucketDataKey } from './models.js';
+import { CompactOptions } from '../BucketStorage.js';
+
+interface CurrentBucketState {
+  /** Bucket name */
+  bucket: string;
+  /**
+   * Rows seen in the bucket, with the last op_id of each.
+   */
+  seen: Map<string, bigint>;
+  /**
+   * Estimated memory usage of the seen Map.
+   */
+  trackingSize: number;
+
+  /**
+   * Last (lowest) seen op_id that is not a PUT.
+   */
+  lastNotPut: bigint | null;
+
+  /**
+   * Number of REMOVE/MOVE operations seen since lastNotPut.
+   */
+  opsSincePut: number;
+}
+
+/**
+ * Additional options, primarily for testing.
+ */
+export interface MongoCompactOptions extends CompactOptions {
+  /** Minimum of 2 */
+  clearBatchLimit?: number;
+  /** Minimum of 1 */
+  moveBatchLimit?: number;
+  /** Minimum of 1 */
+  moveBatchQueryLimit?: number;
+}
+
+const DEFAULT_CLEAR_BATCH_LIMIT = 5000;
+const DEFAULT_MOVE_BATCH_LIMIT = 2000;
+const DEFAULT_MOVE_BATCH_QUERY_LIMIT = 10_000;
+
+/** This default is primarily for tests. */
+const DEFAULT_MEMORY_LIMIT_MB = 64;
+
+export class MongoCompactor {
+  private updates: AnyBulkWriteOperation<BucketDataDocument>[] = [];
+
+  private idLimitBytes: number;
+  private moveBatchLimit: number;
+  private moveBatchQueryLimit: number;
+  private clearBatchLimit: number;
+  private maxOpId: bigint | undefined;
+
+  constructor(private db: PowerSyncMongo, private group_id: number, options?: MongoCompactOptions) {
+    this.idLimitBytes = (options?.memoryLimitMB ?? DEFAULT_MEMORY_LIMIT_MB) * 1024 * 1024;
+    this.moveBatchLimit = options?.moveBatchLimit ?? DEFAULT_MOVE_BATCH_LIMIT;
+    this.moveBatchQueryLimit = options?.moveBatchQueryLimit ?? DEFAULT_MOVE_BATCH_QUERY_LIMIT;
+    this.clearBatchLimit = options?.clearBatchLimit ?? DEFAULT_CLEAR_BATCH_LIMIT;
+    this.maxOpId = options?.maxOpId;
+  }
+
+  /**
+   * Compact buckets by converting operations into MOVE and/or CLEAR operations.
+   *
+   * See /docs/compacting-operations.md for details.
+   */
+  async compact() {
+    const idLimitBytes = this.idLimitBytes;
+
+    let currentState: CurrentBucketState | null = null;
+
+    // Constant lower bound
+    const lowerBound: BucketDataKey = {
+      g: this.group_id,
+      b: new MinKey() as any,
+      o: new MinKey() as any
+    };
+
+    // Upper bound is adjusted for each batch
+    let upperBound: BucketDataKey = {
+      g: this.group_id,
+      b: new MaxKey() as any,
+      o: new MaxKey() as any
+    };
+
+    while (true) {
+      // Query one batch at a time, to avoid cursor timeouts
+      const batch = await this.db.bucket_data
+        .find(
+          {
+            _id: {
+              $gte: lowerBound,
+              $lt: upperBound
+            }
+          },
+          {
+            projection: {
+              _id: 1,
+              op: 1,
+              table: 1,
+              row_id: 1,
+              source_table: 1,
+              source_key: 1
+            },
+            limit: this.moveBatchQueryLimit,
+            sort: { _id: -1 },
+            singleBatch: true
+          }
+        )
+        .toArray();
+
+      if (batch.length == 0) {
+        // We've reached the end
+        break;
+      }
+
+      // Set upperBound for the next batch
+      upperBound = batch[batch.length - 1]._id;
+
+      for (let doc of batch) {
+        if (currentState == null || doc._id.b != currentState.bucket) {
+          if (currentState != null && currentState.lastNotPut != null && currentState.opsSincePut >= 1) {
+            // Important to flush before clearBucket()
+            await this.flush();
+            logger.info(
+              `Inserting CLEAR at ${this.group_id}:${currentState.bucket}:${currentState.lastNotPut} to remove ${currentState.opsSincePut} operations`
+            );
+
+            const bucket = currentState.bucket;
+            const clearOp = currentState.lastNotPut;
+            // Free memory before clearing bucket
+            currentState = null;
+            await this.clearBucket(bucket, clearOp);
+          }
+          currentState = {
+            bucket: doc._id.b,
+            seen: new Map(),
+            trackingSize: 0,
+            lastNotPut: null,
+            opsSincePut: 0
+          };
+        }
+
+        if (this.maxOpId != null && doc._id.o > this.maxOpId) {
+          continue;
+        }
+
+        let isPersistentPut = doc.op == 'PUT';
+
+        if (doc.op == 'REMOVE' || doc.op == 'PUT') {
+          const key = `${doc.table}/${doc.row_id}/${doc.source_table}/${doc.source_key?.toHexString()}`;
+          const targetOp = currentState.seen.get(key);
+          if (targetOp) {
+            // Will convert to MOVE, so don't count as PUT
+            isPersistentPut = false;
+
+            this.updates.push({
+              updateOne: {
+                filter: {
+                  _id: doc._id
+                },
+                update: {
+                  $set: {
+                    op: 'MOVE',
+                    target_op: targetOp
+                  },
+                  $unset: {
+                    source_table: 1,
+                    source_key: 1,
+                    table: 1,
+                    row_id: 1,
+                    data: 1
+                  }
+                }
+              }
+            });
+          } else {
+            if (currentState.trackingSize >= idLimitBytes) {
+              // Reached memory limit.
+              // Keep the highest seen values in this case.
+            } else {
+              // flatstr reduces the memory usage by flattening the string
+              currentState.seen.set(flatstr(key), doc._id.o);
+              // length + 16 for the string
+              // 24 for the bigint
+              // 50 for map overhead
+              // 50 for additional overhead
+              currentState.trackingSize += key.length + 140;
+            }
+          }
+        }
+
+        if (isPersistentPut) {
+          currentState.lastNotPut = null;
+          currentState.opsSincePut = 0;
+        } else if (doc.op != 'CLEAR') {
+          if (currentState.lastNotPut == null) {
+            currentState.lastNotPut = doc._id.o;
+          }
+          currentState.opsSincePut += 1;
+        }
+
+        if (this.updates.length >= this.moveBatchLimit) {
+          await this.flush();
+        }
+      }
+    }
+
+    await this.flush();
+    currentState?.seen.clear();
+    if (currentState?.lastNotPut != null && currentState?.opsSincePut > 1) {
+      logger.info(
+        `Inserting CLEAR at ${this.group_id}:${currentState.bucket}:${currentState.lastNotPut} to remove ${currentState.opsSincePut} operations`
+      );
+      const bucket = currentState.bucket;
+      const clearOp = currentState.lastNotPut;
+      // Free memory before clearing bucket
+      currentState = null;
+      await this.clearBucket(bucket, clearOp);
+    }
+  }
+
+  private async flush() {
+    if (this.updates.length > 0) {
+      logger.info(`Compacting ${this.updates.length} ops`);
+      await this.db.bucket_data.bulkWrite(this.updates, {
+        // Order is not important.
+        // Since checksums are not affected, these operations can happen in any order,
+        // and it's fine if the operations are partially applied.
+        // Each individual operation is atomic.
+        ordered: false
+      });
+      this.updates = [];
+    }
+  }
+
+  /**
+   * Perform a CLEAR compact for a bucket.
+   *
+   * @param bucket bucket name
+   * @param op op_id of the last non-PUT operation, which will be converted to CLEAR.
+   */
+  private async clearBucket(bucket: string, op: bigint) {
+    const opFilter = {
+      _id: {
+        $gte: {
+          g: this.group_id,
+          b: bucket,
+          o: new MinKey() as any
+        },
+        $lte: {
+          g: this.group_id,
+          b: bucket,
+          o: op
+        }
+      }
+    };
+
+    const session = this.db.client.startSession();
+    try {
+      let done = false;
+      while (!done) {
+        // Do the CLEAR operation in batches, with each batch a separate transaction.
+        // The state after each batch is fully consistent.
+        // We need a transaction per batch to make sure checksums stay consistent.
+        await session.withTransaction(
+          async () => {
+            const query = this.db.bucket_data.find(opFilter, {
+              session,
+              sort: { _id: 1 },
+              projection: {
+                _id: 1,
+                op: 1,
+                checksum: 1,
+                target_op: 1
+              },
+              limit: this.clearBatchLimit
+            });
+            let checksum = 0;
+            let lastOpId: BucketDataKey | null = null;
+            let targetOp: bigint | null = null;
+            let gotAnOp = false;
+            for await (let op of query.stream()) {
+              if (op.op == 'MOVE' || op.op == 'REMOVE' || op.op == 'CLEAR') {
+                checksum = addChecksums(checksum, op.checksum);
+                lastOpId = op._id;
+                if (op.op != 'CLEAR') {
+                  gotAnOp = true;
+                }
+                if (op.target_op != null) {
+                  if (targetOp == null || op.target_op > targetOp) {
+                    targetOp = op.target_op;
+                  }
+                }
+              } else {
+                throw new Error(`Unexpected ${op.op} operation at ${op._id.g}:${op._id.b}:${op._id.o}`);
+              }
+            }
+            if (!gotAnOp) {
+              done = true;
+              return;
+            }
+
+            logger.info(`Flushing CLEAR at ${lastOpId?.o}`);
+            await this.db.bucket_data.deleteMany(
+              {
+                _id: {
+                  $gte: {
+                    g: this.group_id,
+                    b: bucket,
+                    o: new MinKey() as any
+                  },
+                  $lte: lastOpId!
+                }
+              },
+              { session }
+            );
+
+            await this.db.bucket_data.insertOne(
+              {
+                _id: lastOpId!,
+                op: 'CLEAR',
+                checksum: checksum,
+                data: null,
+                target_op: targetOp
+              },
+              { session }
+            );
+          },
+          {
+            writeConcern: { w: 'majority' },
+            readConcern: { level: 'snapshot' }
+          }
+        );
+      }
+    } finally {
+      await session.endSession();
+    }
+  }
+}
+
+/**
+ * Flattens string to reduce memory usage (around 320 bytes -> 120 bytes),
+ * at the cost of some upfront CPU usage.
+ *
+ * From: https://github.com/davidmarkclements/flatstr/issues/8
+ */
+function flatstr(s: string) {
+  s.match(/\n/g);
+  return s;
+}

package/src/storage/mongo/MongoSyncBucketStorage.ts

@@ -8,21 +8,24 @@ import * as util from '../../util/util-index.js';
 import {
   BucketDataBatchOptions,
   BucketStorageBatch,
+  CompactOptions,
   DEFAULT_DOCUMENT_BATCH_LIMIT,
   DEFAULT_DOCUMENT_CHUNK_LIMIT_BYTES,
   FlushedResult,
   ResolveTableOptions,
   ResolveTableResult,
+  SyncBucketDataBatch,
   SyncRulesBucketStorage,
   SyncRuleStatus
 } from '../BucketStorage.js';
+import { ChecksumCache, FetchPartialBucketChecksum } from '../ChecksumCache.js';
 import { MongoBucketStorage } from '../MongoBucketStorage.js';
 import { SourceTable } from '../SourceTable.js';
 import { PowerSyncMongo } from './db.js';
 import { BucketDataDocument, BucketDataKey, SourceKey, SyncRuleState } from './models.js';
 import { MongoBucketBatch } from './MongoBucketBatch.js';
-import { BSON_DESERIALIZE_OPTIONS, idPrefixFilter, readSingleBatch, serializeLookup } from './util.js';
-import { ChecksumCache, FetchPartialBucketChecksum } from '../ChecksumCache.js';
+import { MongoCompactor } from './MongoCompactor.js';
+import { BSON_DESERIALIZE_OPTIONS, idPrefixFilter, mapOpEntry, readSingleBatch, serializeLookup } from './util.js';
 
 export class MongoSyncBucketStorage implements SyncRulesBucketStorage {
   private readonly db: PowerSyncMongo;
@@ -201,7 +204,7 @@ export class MongoSyncBucketStorage implements SyncRulesBucketStorage {
     checkpoint: util.OpId,
     dataBuckets: Map<string, string>,
     options?: BucketDataBatchOptions
-  ): AsyncIterable<util.SyncBucketData> {
+  ): AsyncIterable<SyncBucketDataBatch> {
     if (dataBuckets.size == 0) {
       return;
     }
@@ -267,6 +270,7 @@ export class MongoSyncBucketStorage implements SyncRulesBucketStorage {
 
     let batchSize = 0;
     let currentBatch: util.SyncBucketData | null = null;
+    let targetOp: bigint | null = null;
 
     // Ordered by _id, meaning buckets are grouped together
     for (let rawData of data) {
@@ -284,7 +288,8 @@ export class MongoSyncBucketStorage implements SyncRulesBucketStorage {
           start = currentBatch.after;
           currentBatch = null;
           batchSize = 0;
-          yield yieldBatch;
+          yield { batch: yieldBatch, targetOp: targetOp };
+          targetOp = null;
         }
 
         start ??= dataBuckets.get(bucket);
@@ -298,17 +303,18 @@ export class MongoSyncBucketStorage implements SyncRulesBucketStorage {
           data: [],
           next_after: start
         };
+        targetOp = null;
+      }
+
+      const entry = mapOpEntry(row);
+
+      if (row.target_op != null) {
+        // MOVE, CLEAR
+        if (targetOp == null || row.target_op > targetOp) {
+          targetOp = row.target_op;
+        }
       }
 
-      const entry: util.OplogEntry = {
-        op_id: util.timestampToOpId(row._id.o),
-        op: row.op,
-        object_type: row.table,
-        object_id: row.row_id,
-        checksum: Number(row.checksum),
-        subkey: `${row.source_table}/${row.source_key.toHexString()}`,
-        data: row.data
-      };
       currentBatch.data.push(entry);
       currentBatch.next_after = entry.op_id;
 
@@ -318,7 +324,8 @@ export class MongoSyncBucketStorage implements SyncRulesBucketStorage {
     if (currentBatch != null) {
       const yieldBatch = currentBatch;
       currentBatch = null;
-      yield yieldBatch;
+      yield { batch: yieldBatch, targetOp: targetOp };
+      targetOp = null;
     }
   }
 
@@ -530,4 +537,8 @@ export class MongoSyncBucketStorage implements SyncRulesBucketStorage {
       }
     );
   }
+
+  async compact(options?: CompactOptions) {
+    return new MongoCompactor(this.db, this.group_id, options).compact();
+  }
 }

package/src/storage/mongo/models.ts

@@ -42,12 +42,13 @@ export interface BucketParameterDocument {
 export interface BucketDataDocument {
   _id: BucketDataKey;
   op: OpType;
-  source_table: bson.ObjectId;
-  source_key: bson.UUID;
-  table: string;
-  row_id: string;
+  source_table?: bson.ObjectId;
+  source_key?: bson.UUID;
+  table?: string;
+  row_id?: string;
   checksum: number;
   data: string | null;
+  target_op?: bigint | null;
 }
 
 export type OpType = 'PUT' | 'REMOVE' | 'MOVE' | 'CLEAR';

package/src/storage/mongo/util.ts

@@ -2,6 +2,9 @@ import { SqliteJsonValue } from '@powersync/service-sync-rules';
 import * as bson from 'bson';
 import * as mongo from 'mongodb';
 import * as crypto from 'crypto';
+import { BucketDataDocument } from './models.js';
+import { timestampToOpId } from '../../util/utils.js';
+import { OplogEntry } from '../../util/protocol-types.js';
 
 /**
  * Lookup serialization must be number-agnostic. I.e. normalize numbers, instead of preserving numbers.
@@ -86,3 +89,25 @@ export const BSON_DESERIALIZE_OPTIONS: bson.DeserializeOptions = {
   // use bigint instead of Long
   useBigInt64: true
 };
+
+export function mapOpEntry(row: BucketDataDocument): OplogEntry {
+  if (row.op == 'PUT' || row.op == 'REMOVE') {
+    return {
+      op_id: timestampToOpId(row._id.o),
+      op: row.op,
+      object_type: row.table,
+      object_id: row.row_id,
+      checksum: Number(row.checksum),
+      subkey: `${row.source_table}/${row.source_key!.toHexString()}`,
+      data: row.data
+    };
+  } else {
+    // MOVE, CLEAR
+
+    return {
+      op_id: timestampToOpId(row._id.o),
+      op: row.op,
+      checksum: Number(row.checksum)
+    };
+  }
+}

package/src/sync/RequestTracker.ts

@@ -0,0 +1,21 @@
+import { Metrics } from '../metrics/Metrics.js';
+
+/**
+ * Record sync stats per request stream.
+ */
+export class RequestTracker {
+  operationsSynced = 0;
+  dataSyncedBytes = 0;
+
+  addOperationsSynced(operations: number) {
+    this.operationsSynced += operations;
+
+    Metrics.getInstance().operations_synced_total.add(operations);
+  }
+
+  addDataSynced(bytes: number) {
+    this.dataSyncedBytes += bytes;
+
+    Metrics.getInstance().data_synced_bytes.add(bytes);
+  }
+}