@powersync/service-module-mongodb 0.0.0-dev-20241128134723 → 0.0.0-dev-20241219110735

package/src/storage/implementation/MongoCompactor.ts

@@ -0,0 +1,392 @@
+import { logger } from '@powersync/lib-services-framework';
+import { storage, utils } from '@powersync/service-core';
+import { AnyBulkWriteOperation, MaxKey, MinKey } from 'mongodb';
+import { PowerSyncMongo } from './db.js';
+import { BucketDataDocument, BucketDataKey } from './models.js';
+import { cacheKey } from './OperationBatch.js';
+import { safeBulkWrite } from './util.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 storage.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;
+  private buckets: string[] | 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;
+    this.buckets = options?.compactBuckets;
+  }
+
+  /**
+   * Compact buckets by converting operations into MOVE and/or CLEAR operations.
+   *
+   * See /docs/compacting-operations.md for details.
+   */
+  async compact() {
+    if (this.buckets) {
+      for (let bucket of this.buckets) {
+        // We can make this more efficient later on by iterating
+        // through the buckets in a single query.
+        // That makes batching more tricky, so we leave for later.
+        await this.compactInternal(bucket);
+      }
+    } else {
+      await this.compactInternal(undefined);
+    }
+  }
+
+  async compactInternal(bucket: string | undefined) {
+    const idLimitBytes = this.idLimitBytes;
+
+    let currentState: CurrentBucketState | null = null;
+
+    let bucketLower: string | MinKey;
+    let bucketUpper: string | MaxKey;
+
+    if (bucket == null) {
+      bucketLower = new MinKey();
+      bucketUpper = new MaxKey();
+    } else if (bucket.includes('[')) {
+      // Exact bucket name
+      bucketLower = bucket;
+      bucketUpper = bucket;
+    } else {
+      // Bucket definition name
+      bucketLower = `${bucket}[`;
+      bucketUpper = `${bucket}[\uFFFF`;
+    }
+
+    // Constant lower bound
+    const lowerBound: BucketDataKey = {
+      g: this.group_id,
+      b: bucketLower as string,
+      o: new MinKey() as any
+    };
+
+    // Upper bound is adjusted for each batch
+    let upperBound: BucketDataKey = {
+      g: this.group_id,
+      b: bucketUpper as string,
+      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}/${doc.row_id}/${cacheKey(doc.source_table!, doc.source_key!)}`;
+          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 safeBulkWrite(this.db.bucket_data, 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 = utils.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/implementation/MongoIdSequence.ts

@@ -0,0 +1,24 @@
+/**
+ * Manages op_id or similar sequence in memory.
+ *
+ * This is typically used within a transaction, with the last value persisted
+ * at the end of the transaction.
+ */
+export class MongoIdSequence {
+  private _last: bigint;
+
+  constructor(last: bigint) {
+    if (typeof last != 'bigint') {
+      throw new Error(`BigInt required, got ${last} ${typeof last}`);
+    }
+    this._last = last;
+  }
+
+  next() {
+    return ++this._last;
+  }
+
+  last() {
+    return this._last;
+  }
+}

package/src/storage/implementation/MongoPersistedSyncRules.ts

@@ -0,0 +1,16 @@
+import { SqlSyncRules } from '@powersync/service-sync-rules';
+
+import { storage } from '@powersync/service-core';
+
+export class MongoPersistedSyncRules implements storage.PersistedSyncRules {
+  public readonly slot_name: string;
+
+  constructor(
+    public readonly id: number,
+    public readonly sync_rules: SqlSyncRules,
+    public readonly checkpoint_lsn: string | null,
+    slot_name: string | null
+  ) {
+    this.slot_name = slot_name ?? `powersync_${id}`;
+  }
+}

package/src/storage/implementation/MongoPersistedSyncRulesContent.ts

@@ -0,0 +1,49 @@
+import { storage } from '@powersync/service-core';
+import { SqlSyncRules } from '@powersync/service-sync-rules';
+import * as mongo from 'mongodb';
+import { MongoPersistedSyncRules } from './MongoPersistedSyncRules.js';
+import { MongoSyncRulesLock } from './MongoSyncRulesLock.js';
+import { PowerSyncMongo } from './db.js';
+import { SyncRuleDocument } from './models.js';
+
+export class MongoPersistedSyncRulesContent implements storage.PersistedSyncRulesContent {
+  public readonly slot_name: string;
+
+  public readonly id: number;
+  public readonly sync_rules_content: string;
+  public readonly last_checkpoint_lsn: string | null;
+  public readonly last_fatal_error: string | null;
+  public readonly last_keepalive_ts: Date | null;
+  public readonly last_checkpoint_ts: Date | null;
+
+  public current_lock: MongoSyncRulesLock | null = null;
+
+  constructor(
+    private db: PowerSyncMongo,
+    doc: mongo.WithId<SyncRuleDocument>
+  ) {
+    this.id = doc._id;
+    this.sync_rules_content = doc.content;
+    this.last_checkpoint_lsn = doc.last_checkpoint_lsn;
+    // Handle legacy values
+    this.slot_name = doc.slot_name ?? `powersync_${this.id}`;
+    this.last_fatal_error = doc.last_fatal_error;
+    this.last_checkpoint_ts = doc.last_checkpoint_ts;
+    this.last_keepalive_ts = doc.last_keepalive_ts;
+  }
+
+  parsed(options: storage.ParseSyncRulesOptions) {
+    return new MongoPersistedSyncRules(
+      this.id,
+      SqlSyncRules.fromYaml(this.sync_rules_content, options),
+      this.last_checkpoint_lsn,
+      this.slot_name
+    );
+  }
+
+  async lock() {
+    const lock = await MongoSyncRulesLock.createLock(this.db, this);
+    this.current_lock = lock;
+    return lock;
+  }
+}

package/src/storage/implementation/MongoStorageProvider.ts

@@ -0,0 +1,42 @@
+import { logger } from '@powersync/lib-services-framework';
+import { storage } from '@powersync/service-core';
+import { configFile } from '@powersync/service-types';
+import * as db from '../../db/db-index.js';
+import { MongoBucketStorage } from '../MongoBucketStorage.js';
+import { PowerSyncMongo } from './db.js';
+
+export const MONGO_STORAGE_TYPE = 'mongodb';
+export class MongoStorageProvider implements storage.BucketStorageProvider {
+  get type() {
+    return MONGO_STORAGE_TYPE;
+  }
+
+  async getStorage(options: storage.GetStorageOptions): Promise<storage.ActiveStorage> {
+    const { resolvedConfig } = options;
+
+    const { storage } = resolvedConfig;
+    if (storage.type != MONGO_STORAGE_TYPE) {
+      // This should not be reached since the generation should be managed externally.
+      throw new Error(
+        `Cannot create MongoDB bucket storage with provided config ${storage.type} !== ${MONGO_STORAGE_TYPE}`
+      );
+    }
+
+    const decodedConfig = configFile.MongoStorageConfig.decode(storage as any);
+    const client = db.mongo.createMongoClient(decodedConfig);
+
+    const database = new PowerSyncMongo(client, { database: resolvedConfig.storage.database });
+
+    return {
+      storage: new MongoBucketStorage(database, {
+        // TODO currently need the entire resolved config due to this
+        slot_name_prefix: resolvedConfig.slot_name_prefix
+      }),
+      shutDown: () => client.close(),
+      tearDown: () => {
+        logger.info(`Tearing down storage: ${database.db.namespace}...`);
+        return database.db.dropDatabase();
+      }
+    } satisfies storage.ActiveStorage;
+  }
+}