@fluidframework/container-runtime 1.2.6 → 2.0.0-dev.1.3.0.96595

package/dist/blobManager.js

@@ -5,10 +5,12 @@
  */
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.BlobManager = exports.BlobHandle = void 0;
+const uuid_1 = require("uuid");
 const runtime_utils_1 = require("@fluidframework/runtime-utils");
 const common_utils_1 = require("@fluidframework/common-utils");
 const container_definitions_1 = require("@fluidframework/container-definitions");
 const telemetry_utils_1 = require("@fluidframework/telemetry-utils");
+const throttler_1 = require("./throttler");
 /**
  * This class represents blob (long string)
  * This object is used only when creating (writing) new blob and serialization purposes.
@@ -36,33 +38,111 @@ class BlobHandle {
     }
 }
 exports.BlobHandle = BlobHandle;
+class CancellableThrottler {
+    constructor(throttler) {
+        this.throttler = throttler;
+        this.cancelP = new common_utils_1.Deferred();
+    }
+    async getDelay() {
+        return Promise.race([
+            this.cancelP.promise,
+            new Promise((resolve) => setTimeout(resolve, this.throttler.getDelay())),
+        ]);
+    }
+    cancel() {
+        this.cancelP.resolve();
+        this.cancelP = new common_utils_1.Deferred();
+    }
+}
+// Note that while offline we "submit" an op before uploading the blob, but we always
+// expect blobs to be uploaded before we actually see the op round-trip
+var PendingBlobStatus;
+(function (PendingBlobStatus) {
+    PendingBlobStatus[PendingBlobStatus["OnlinePendingUpload"] = 0] = "OnlinePendingUpload";
+    PendingBlobStatus[PendingBlobStatus["OnlinePendingOp"] = 1] = "OnlinePendingOp";
+    PendingBlobStatus[PendingBlobStatus["OfflinePendingUpload"] = 2] = "OfflinePendingUpload";
+    PendingBlobStatus[PendingBlobStatus["OfflinePendingOp"] = 3] = "OfflinePendingOp";
+})(PendingBlobStatus || (PendingBlobStatus = {}));
 class BlobManager {
-    constructor(routeContext, snapshot, getStorage, attachBlobCallback, 
+    constructor(routeContext, snapshot, getStorage, 
+    /**
+     * Submit a BlobAttach op. When a blob is uploaded, there is a short grace period before which
+     * the blob is deleted. The BlobAttach op notifies the server that blob is in use. The server
+     * will then not delete the blob as long as it is listed as referenced in future summaries.
+     * The summarizing client will know to include the storage ID in the summary when it sees the op.
+     *
+     * The op may also include a local ID to inform all clients of the relation to the storage
+     * ID, without knowledge of which they cannot request the blob from storage. This is also
+     * included in the redirect table in the summary.
+     */
+    sendBlobAttachOp, 
     // To be called when a blob node is requested. blobPath is the path of the blob's node in GC's graph. It's
     // of the format `/<BlobManager.basePath>/<blobId>`.
-    gcNodeUpdated, runtime, logger) {
+    gcNodeUpdated, runtime, stashedBlobs = {}) {
         this.routeContext = routeContext;
         this.getStorage = getStorage;
-        this.attachBlobCallback = attachBlobCallback;
+        this.sendBlobAttachOp = sendBlobAttachOp;
         this.gcNodeUpdated = gcNodeUpdated;
         this.runtime = runtime;
-        this.logger = logger;
-        // uploaded blob IDs
-        this.blobIds = new Set();
-        // blobs for which upload is pending. maps to a promise that will resolve once the blob has been uploaded and a
-        // BlobAttach op has round-tripped.
-        this.pendingBlobIds = new Map();
-        // blobs uploaded while detached; cleared upon attach
-        this.detachedBlobIds = new Set();
-        this.runtime.once("dispose", () => {
-            for (const promise of this.pendingBlobIds.values()) {
-                promise.reject(new Error("runtime disposed while blobAttach op in flight"));
-            }
+        /**
+         * Blobs which have not been uploaded or for which we have not yet seen a BlobAttach op round-trip.
+         * Until we see the op round-trip, there is a possibility we may need to re-upload the blob, so
+         * we must save it. This is true for both the online and offline flow.
+         */
+        this.pendingBlobs = new Map();
+        /**
+         * Track ops in flight for online flow. Used to avoid searching pendingBlobs since BlobAttach ops
+         * don't include local ID in online flow.
+         */
+        this.opsInFlight = new Map();
+        this.retryThrottler = new CancellableThrottler(new throttler_1.Throttler(60 * 1000, // 60 sec delay window
+        30 * 1000, // 30 sec max delay
+        // throttling function increases exponentially (0ms, 40ms, 80ms, 160ms, etc)
+        (0, throttler_1.formExponentialFn)({ coefficient: 20, initialDelay: 0 })));
+        this.logger = telemetry_utils_1.ChildLogger.create(this.runtime.logger, "BlobManager");
+        this.runtime.on("disconnected", () => this.onDisconnected());
+        this.redirectTable = this.load(snapshot);
+        // Begin uploading stashed blobs from previous container instance
+        Object.entries(stashedBlobs).forEach(([localId, entry]) => {
+            const blob = (0, common_utils_1.stringToBuffer)(entry.blob, "base64");
+            this.pendingBlobs.set(localId, {
+                blob,
+                status: PendingBlobStatus.OfflinePendingUpload,
+                handleP: new common_utils_1.Deferred(),
+                uploadP: this.uploadBlob(localId, blob),
+            });
         });
-        this.load(snapshot);
     }
-    hasBlob(id) {
-        return this.blobIds.has(id) || this.detachedBlobIds.has(id);
+    get pendingOfflineUploads() {
+        return Array.from(this.pendingBlobs.values())
+            .filter((e) => e.status === PendingBlobStatus.OfflinePendingUpload);
+    }
+    get hasPendingOfflineUploads() {
+        return this.pendingOfflineUploads.length > 0;
+    }
+    /**
+     * Upload blobs added while offline. This must be completed before connecting and resubmitting ops.
+     */
+    async onConnected() {
+        this.retryThrottler.cancel();
+        const pendingUploads = this.pendingOfflineUploads.map(async (e) => e.uploadP);
+        await telemetry_utils_1.PerformanceEvent.timedExecAsync(this.logger, {
+            eventName: "BlobUploadOnConnected",
+            count: pendingUploads.length,
+        }, async () => Promise.all(pendingUploads), { start: true, end: true });
+    }
+    /**
+     * Transition online blobs waiting for BlobAttach op round-trip since we will not see the op until we are connected
+     * again
+     */
+    onDisconnected() {
+        for (const [localId, entry] of this.pendingBlobs) {
+            if (entry.status === PendingBlobStatus.OnlinePendingOp) {
+                // This will submit another BlobAttach op for this blob. This is necessary because the one we sent
+                // already didn't have the local ID.
+                this.transitionToOffline(localId);
+            }
+        }
     }
     /**
      * For a blobId, returns its path in GC's graph. The node path is of the format `/<BlobManager.basePath>/<blobId>`
@@ -72,61 +152,200 @@ class BlobManager {
     getBlobGCNodePath(blobId) {
         return `/${BlobManager.basePath}/${blobId}`;
     }
+    /**
+     * Set of actual storage IDs (i.e., IDs that can be requested from storage). This will be empty if the container is
+     * detached or there are no (non-pending) attachment blobs in the document
+     */
+    get storageIds() {
+        const ids = new Set(this.redirectTable.values());
+        // If we are detached, we will not have storage IDs, only undefined
+        const undefinedValueInTable = ids.delete(undefined);
+        // For a detached container, entries are inserted into the redirect table with an undefined storage ID.
+        // For an attached container, entries are inserted w/storage ID after the BlobAttach op round-trips.
+        (0, common_utils_1.assert)(!undefinedValueInTable || this.runtime.attachState === container_definitions_1.AttachState.Detached && ids.size === 0, 0x382 /* 'redirectTable' must contain only undefined while detached / defined values while attached */);
+        return ids;
+    }
     async getBlob(blobId) {
-        var _a, _b;
-        const storageId = (_b = (_a = this.redirectTable) === null || _a === void 0 ? void 0 : _a.get(blobId)) !== null && _b !== void 0 ? _b : blobId;
-        (0, common_utils_1.assert)(this.hasBlob(storageId), 0x11f /* "requesting unknown blobs" */);
-        // When this blob is retrieved, let the container runtime know that the corresponding GC node got updated.
+        const pending = this.pendingBlobs.get(blobId);
+        if (pending) {
+            return pending.blob;
+        }
+        let storageId;
+        if (this.runtime.attachState === container_definitions_1.AttachState.Detached) {
+            (0, common_utils_1.assert)(this.redirectTable.has(blobId), 0x383 /* requesting unknown blobs */);
+            // Blobs created while the container is detached are stored in IDetachedBlobStorage.
+            // The 'IDocumentStorageService.readBlob()' call below will retrieve these via localId.
+            storageId = blobId;
+        }
+        else {
+            storageId = this.redirectTable.get(blobId);
+            (0, common_utils_1.assert)(!!storageId, 0x11f /* "requesting unknown blobs" */);
+        }
+        // When a GC-able (not pending) blob is retrieved, let runtime know that the corresponding GC node got updated.
         this.gcNodeUpdated(this.getBlobGCNodePath(blobId));
-        return new BlobHandle(`${BlobManager.basePath}/${storageId}`, this.routeContext, async () => telemetry_utils_1.PerformanceEvent.timedExecAsync(this.logger, { eventName: "AttachmentReadBlob", id: storageId }, async () => {
+        return telemetry_utils_1.PerformanceEvent.timedExecAsync(this.logger, { eventName: "AttachmentReadBlob", id: storageId }, async () => {
             return this.getStorage().readBlob(storageId);
-        }, { end: true, cancel: "error" }));
+        }, { end: true, cancel: "error" });
+    }
+    getBlobHandle(id) {
+        (0, common_utils_1.assert)(this.redirectTable.has(id) || this.pendingBlobs.has(id), 0x384 /* requesting handle for unknown blob */);
+        return new BlobHandle(`${BlobManager.basePath}/${id}`, this.routeContext, async () => this.getBlob(id));
+    }
+    async createBlobDetached(blob) {
+        // Blobs created while the container is detached are stored in IDetachedBlobStorage.
+        // The 'IDocumentStorageService.createBlob()' call below will respond with a localId.
+        const response = await this.getStorage().createBlob(blob);
+        this.redirectTable.set(response.id, undefined);
+        return this.getBlobHandle(response.id);
     }
     async createBlob(blob) {
-        var _a, _b;
+        if (this.runtime.attachState === container_definitions_1.AttachState.Detached) {
+            return this.createBlobDetached(blob);
+        }
         if (this.runtime.attachState === container_definitions_1.AttachState.Attaching) {
             // blob upload is not supported in "Attaching" state
             this.logger.sendTelemetryEvent({ eventName: "CreateBlobWhileAttaching" });
             await new Promise((resolve) => this.runtime.once("attached", resolve));
         }
-        if (!this.runtime.connected && this.runtime.attachState === container_definitions_1.AttachState.Attached) {
-            // see https://github.com/microsoft/FluidFramework/issues/8246
-            // Avoid getting storage if we are offline since it might be undefined. In the future we will return
-            // handles immediately while offline
-            await new Promise((resolve) => this.runtime.once("connected", resolve));
-        }
-        const response = await telemetry_utils_1.PerformanceEvent.timedExecAsync(this.logger, { eventName: "createBlob" }, async () => this.getStorage().createBlob(blob), { end: true, cancel: "error" });
-        const handle = new BlobHandle(`${BlobManager.basePath}/${response.id}`, this.routeContext, 
-        // get() should go through BlobManager.getBlob() so handles created while detached can be redirected
-        // to the correct storage id after they are uploaded
-        async () => this.getBlob(response.id).then(async (h) => h.get()));
-        if (this.runtime.attachState === container_definitions_1.AttachState.Detached) {
-            this.detachedBlobIds.add(response.id);
-            return handle;
+        (0, common_utils_1.assert)(this.runtime.attachState === container_definitions_1.AttachState.Attached, 0x385 /* For clarity and paranoid defense against adding future attachment states */);
+        // Create a local ID for each blob. This is used to support blobs if/when the client goes
+        // offline since we don't have the ID from storage yet. If online flow succeeds this won't be used.
+        const localId = (0, uuid_1.v4)();
+        const pendingEntry = {
+            blob,
+            status: PendingBlobStatus.OnlinePendingUpload,
+            handleP: new common_utils_1.Deferred(),
+            uploadP: this.uploadBlob(localId, blob),
+        };
+        this.pendingBlobs.set(localId, pendingEntry);
+        return pendingEntry.handleP.promise;
+    }
+    async uploadBlob(localId, blob) {
+        return telemetry_utils_1.PerformanceEvent.timedExecAsync(this.logger, { eventName: "createBlob" }, async () => this.getStorage().createBlob(blob), { end: true, cancel: this.runtime.connected ? "error" : "generic" }).then((response) => this.onUploadResolve(localId, response), async (err) => this.onUploadReject(localId, err));
+    }
+    onUploadResolve(localId, response) {
+        var _a;
+        const entry = this.pendingBlobs.get(localId);
+        (0, common_utils_1.assert)((entry === null || entry === void 0 ? void 0 : entry.status) === PendingBlobStatus.OnlinePendingUpload ||
+            (entry === null || entry === void 0 ? void 0 : entry.status) === PendingBlobStatus.OfflinePendingUpload, 0x386 /* Must have pending blob entry for uploaded blob */);
+        entry.storageId = response.id;
+        if (this.runtime.connected) {
+            if (entry.status === PendingBlobStatus.OnlinePendingUpload) {
+                if (this.storageIds.has(response.id)) {
+                    // Storage may dedupe blobs and give us an ID we already know
+                    // no need to submit BlobAttach op in this case
+                    entry.handleP.resolve(this.getBlobHandle(response.id));
+                    this.pendingBlobs.delete(localId);
+                }
+                else {
+                    // Check for still-pending duplicates too; if an op is already in flight we can wait for that one
+                    if (!this.opsInFlight.has(response.id)) {
+                        this.sendBlobAttachOp(response.id);
+                    }
+                    this.opsInFlight.set(response.id, ((_a = this.opsInFlight.get(response.id)) !== null && _a !== void 0 ? _a : []).concat(localId));
+                    entry.status = PendingBlobStatus.OnlinePendingOp;
+                }
+            }
+            else if (entry.status === PendingBlobStatus.OfflinePendingUpload) {
+                // We already submitted a BlobAttach op for this blob when it was transitioned to offline flow
+                entry.status = PendingBlobStatus.OfflinePendingOp;
+            }
         }
-        // Note - server will de-dup blobs, so we might get existing blobId!
-        if (this.pendingBlobIds.has(response.id)) {
-            await ((_a = this.pendingBlobIds.get(response.id)) === null || _a === void 0 ? void 0 : _a.promise);
+        else {
+            // connected to storage but not ordering service?
+            this.logger.sendTelemetryEvent({ eventName: "BlobUploadSuccessWhileDisconnected" });
+            if (entry.status === PendingBlobStatus.OnlinePendingUpload) {
+                this.transitionToOffline(localId);
+            }
+            entry.status = PendingBlobStatus.OfflinePendingOp;
         }
-        else if (!this.blobIds.has(response.id)) {
-            this.pendingBlobIds.set(response.id, new common_utils_1.Deferred());
-            // send blob attach op and wait until we see it to return the handle
-            this.attachBlobCallback(response.id);
-            await ((_b = this.pendingBlobIds.get(response.id)) === null || _b === void 0 ? void 0 : _b.promise);
+        return response;
+    }
+    async onUploadReject(localId, error) {
+        const entry = this.pendingBlobs.get(localId);
+        (0, common_utils_1.assert)(!!entry, 0x387 /* Must have pending blob entry for blob which failed to upload */);
+        if (!this.runtime.connected) {
+            if (entry.status === PendingBlobStatus.OnlinePendingUpload) {
+                this.transitionToOffline(localId);
+            }
+            // we are probably not connected to storage but start another upload request in case we are
+            entry.uploadP = this.retryThrottler.getDelay().then(async () => this.uploadBlob(localId, entry.blob));
+            return entry.uploadP;
         }
-        return handle;
+        else {
+            entry.handleP.reject(error);
+            throw error;
+        }
+    }
+    transitionToOffline(localId) {
+        (0, common_utils_1.assert)(!this.runtime.connected, 0x388 /* Must only transition to offline flow while runtime is disconnected */);
+        const entry = this.pendingBlobs.get(localId);
+        (0, common_utils_1.assert)(!!entry, 0x389 /* No pending blob entry */);
+        (0, common_utils_1.assert)([PendingBlobStatus.OnlinePendingUpload, PendingBlobStatus.OnlinePendingOp].includes(entry.status), 0x38a /* Blob must be in online flow to transition to offline flow */);
+        entry.status = entry.status === PendingBlobStatus.OnlinePendingUpload
+            ? PendingBlobStatus.OfflinePendingUpload
+            : PendingBlobStatus.OfflinePendingOp;
+        // Submit a BlobAttach op. It's important we submit this op now before returning the blob handle so the
+        // BlobAttach op is sequenced prior to any ops referencing the handle. Otherwise an invalid handle could be
+        // added to the document if the ops are not all successfully submitted upon reconnection.
+        // storageId may be undefined but since we are not connected we will have a chance to add it when reSubmit()
+        // is called
+        this.sendBlobAttachOp(entry.storageId, localId);
+        entry.handleP.resolve(this.getBlobHandle(localId));
+    }
+    /**
+     * Resubmit a BlobAttach op. Used to add storage IDs to ops that were
+     * submitted to runtime while disconnected.
+     * @param metadata - op metadata containing storage and/or local IDs
+     */
+    reSubmit(metadata) {
+        (0, common_utils_1.assert)(!!metadata, 0x38b /* Resubmitted ops must have metadata */);
+        const { blobId, localId } = metadata;
+        if (!blobId) {
+            (0, common_utils_1.assert)(!!localId, 0x38c /* Submitted BlobAttach ops must have a blobId or localId */);
+            // We submitted this op while offline. The blob should have been uploaded by now.
+            const pendingEntry = this.pendingBlobs.get(localId);
+            (0, common_utils_1.assert)((pendingEntry === null || pendingEntry === void 0 ? void 0 : pendingEntry.status) === PendingBlobStatus.OfflinePendingOp &&
+                !!(pendingEntry === null || pendingEntry === void 0 ? void 0 : pendingEntry.storageId), 0x38d /* blob must be uploaded before resubmitting BlobAttach op */);
+            return this.sendBlobAttachOp(pendingEntry.storageId, localId);
+        }
+        return this.sendBlobAttachOp(blobId, localId);
     }
-    processBlobAttachOp(blobId, local) {
+    processBlobAttachOp(message, local) {
+        var _a, _b;
+        (0, common_utils_1.assert)((_a = message === null || message === void 0 ? void 0 : message.metadata) === null || _a === void 0 ? void 0 : _a.blobId, 0x12a /* "Missing blob id on metadata" */);
+        if (message.metadata.localId !== undefined) {
+            this.redirectTable.set(message.metadata.localId, message.metadata.blobId);
+        }
+        // set identity (id -> id) entry
+        this.redirectTable.set(message.metadata.blobId, message.metadata.blobId);
         if (local) {
-            const pendingBlobP = this.pendingBlobIds.get(blobId);
-            (0, common_utils_1.assert)(pendingBlobP !== undefined, 0x1f8 /* "local BlobAttach op with no pending blob" */);
-            pendingBlobP.resolve();
-            this.pendingBlobIds.delete(blobId);
+            if (message.metadata.localId === undefined) {
+                // Since there is no local ID, we know this op was submitted while online.
+                const waitingBlobs = this.opsInFlight.get(message.metadata.blobId);
+                (0, common_utils_1.assert)(!!waitingBlobs, 0x38e /* local online BlobAttach op with no pending blob */);
+                waitingBlobs.forEach((localId) => {
+                    const pendingBlobEntry = this.pendingBlobs.get(localId);
+                    (0, common_utils_1.assert)(pendingBlobEntry !== undefined, 0x38f);
+                    // It's possible we transitioned to offline flow while waiting for this op.
+                    if (pendingBlobEntry.status === PendingBlobStatus.OnlinePendingOp) {
+                        pendingBlobEntry.handleP.resolve(this.getBlobHandle(message.metadata.blobId));
+                        this.pendingBlobs.delete(localId);
+                    }
+                });
+            }
+            else {
+                // Each local ID is unique; get the pending blob entry and delete it
+                (0, common_utils_1.assert)(((_b = this.pendingBlobs.get(message.metadata.localId)) === null || _b === void 0 ? void 0 : _b.status) === PendingBlobStatus.OfflinePendingOp, 0x1f8 /* "local BlobAttach op with no pending blob" */);
+                this.pendingBlobs.delete(message.metadata.localId);
+            }
         }
-        this.blobIds.add(blobId);
     }
     /**
      * Reads blobs needed to load BlobManager from storage.
+     * @param blobsTree - Tree containing IDs of previously attached blobs. We
+     * look for the IDs in the blob entries of the tree since the both the r11s
+     * and SPO drivers replace the attachment types returned in snapshot() with blobs.
      */
     static async load(blobsTree, tryFetchBlob) {
         if (!blobsTree) {
@@ -142,33 +361,23 @@ class BlobManager {
         return { ids, redirectTable };
     }
     /**
-     * Load a set of previously attached blob IDs from a previous snapshot. Note
-     * that BlobManager tracking and reporting attached blobs is a temporary
-     * solution since storage expects attached blobs to be reported and any that
-     * are not reported as attached may be GCed. In the future attached blob
-     * IDs will be collected at summarization time, and runtime will not care
-     * about the existence or specific formatting of this tree in returned
-     * snapshots.
-     *
-     * @param blobsTree - Tree containing IDs of previously attached blobs. This
-     * corresponds to snapshot() below. We look for the IDs in the blob entries
-     * of the tree since the both the r11s and SPO drivers replace the
-     * attachment types returned in snapshot() with blobs.
+     * Load a set of previously attached blob IDs and redirect table from a previous snapshot.
      */
     load(snapshot) {
         var _a, _b, _c;
-        if (snapshot.ids) {
-            const detached = this.runtime.attachState === container_definitions_1.AttachState.Detached;
-            snapshot.ids.map((entry) => detached ? this.detachedBlobIds.add(entry) : this.blobIds.add(entry));
-        }
-        if (snapshot.redirectTable) {
-            this.redirectTable = new Map(snapshot.redirectTable);
-        }
         this.logger.sendTelemetryEvent({
             eventName: "AttachmentBlobsLoaded",
             count: (_b = (_a = snapshot.ids) === null || _a === void 0 ? void 0 : _a.length) !== null && _b !== void 0 ? _b : 0,
             redirectTable: (_c = snapshot.redirectTable) === null || _c === void 0 ? void 0 : _c.length,
         });
+        const table = new Map(snapshot.redirectTable);
+        if (snapshot.ids) {
+            const detached = this.runtime.attachState === container_definitions_1.AttachState.Detached;
+            // If we are detached, we don't have storage IDs yet, so set to undefined
+            // Otherwise, set identity (id -> id) entries
+            snapshot.ids.forEach((entry) => table.set(entry, detached ? undefined : entry));
+        }
+        return table;
     }
     /**
      * Generates data used for garbage collection. Each blob uploaded represents a node in the GC graph as it can be
@@ -178,22 +387,23 @@ class BlobManager {
      */
     getGCData(fullGC = false) {
         const gcData = { gcNodes: {} };
-        this.blobIds.forEach((blobId) => {
+        /**
+          * The node path is of the format `/_blobs/blobId`. This path must match the path of the blob handle returned
+          * by the createBlob API because blobs are marked referenced by storing these handles in a referenced DDS.
+          */
+        this.storageIds.forEach((blobId) => {
             gcData.gcNodes[this.getBlobGCNodePath(blobId)] = [];
         });
-        /**
-         * For all blobs in the redirect table, the handle returned on creation is based off of the localId. So, these
-         * nodes can be referenced by storing the localId handle. When that happens, the corresponding storageId node
-         * must also be marked referenced. So, we add a route from the localId node to the storageId node.
-         * Note that because of de-duping, there can be multiple localIds that all redirect to the same storageId or
-         * a blob may be referenced via its storageId handle.
-         */
-        if (this.redirectTable !== undefined) {
-            for (const [localId, storageId] of this.redirectTable) {
-                // Add node for the localId and add a route to the storageId node. The storageId node will have been
-                // added above when adding nodes for this.blobIds.
-                gcData.gcNodes[this.getBlobGCNodePath(localId)] = [this.getBlobGCNodePath(storageId)];
-            }
+        // For some blobs, the handle returned on creation is based off of the localId. So, these
+        // nodes can be referenced by storing the localId handle. When that happens, the corresponding storageId node
+        // must also be marked referenced. So, we add a route from the localId node to the storageId node.
+        // Note that because of de-duping, there can be multiple localIds that all redirect to the same storageId or
+        // a blob may be referenced via its storageId handle.
+        for (const [localId, storageId] of this.redirectTable) {
+            (0, common_utils_1.assert)(!!storageId, 0x390 /* Must be attached to get GC data */);
+            // Add node for the localId and add a route to the storageId node. The storageId node will have been
+            // added above when adding nodes for this.blobIds.
+            gcData.gcNodes[this.getBlobGCNodePath(localId)] = [this.getBlobGCNodePath(storageId)];
         }
         return gcData;
     }
@@ -215,32 +425,40 @@ class BlobManager {
                 this.redirectTable.delete(blobId);
                 continue;
             }
-            this.blobIds.delete(blobId);
         }
     }
     summarize(telemetryContext) {
-        // If we have a redirect table it means the container is about to transition to "Attaching" state, so we need
-        // to return an actual snapshot containing all the real storage IDs we know about.
-        const attachingOrAttached = !!this.redirectTable || this.runtime.attachState !== container_definitions_1.AttachState.Detached;
-        const blobIds = attachingOrAttached ? this.blobIds : this.detachedBlobIds;
+        // if storageIds is empty, it means we are detached and have only local IDs, or that there are no blobs attached
+        const blobIds = this.storageIds.size > 0 ? Array.from(this.storageIds) : Array.from(this.redirectTable.keys());
         const builder = new runtime_utils_1.SummaryTreeBuilder();
         blobIds.forEach((blobId) => {
             builder.addAttachment(blobId);
         });
-        if (this.redirectTable && this.redirectTable.size > 0) {
-            builder.addBlob(BlobManager.redirectTableBlobName, JSON.stringify(Array.from(this.redirectTable.entries())));
+        // Any non-identity entries in the table need to be saved in the summary
+        if (this.redirectTable.size > blobIds.length) {
+            builder.addBlob(BlobManager.redirectTableBlobName, 
+            // filter out identity entries
+            JSON.stringify(Array.from(this.redirectTable.entries())
+                .filter(([localId, storageId]) => localId !== storageId)));
         }
         return builder.getSummaryTree();
     }
     setRedirectTable(table) {
         (0, common_utils_1.assert)(this.runtime.attachState === container_definitions_1.AttachState.Detached, 0x252 /* "redirect table can only be set in detached container" */);
-        (0, common_utils_1.assert)(!this.redirectTable, 0x253 /* "redirect table already exists" */);
+        (0, common_utils_1.assert)(this.redirectTable.size === table.size, 0x391 /* Redirect table size must match BlobManager's local ID count */);
         for (const [localId, storageId] of table) {
-            (0, common_utils_1.assert)(this.detachedBlobIds.delete(localId), 0x254 /* "unrecognized id in redirect table" */);
-            this.blobIds.add(storageId);
+            (0, common_utils_1.assert)(this.redirectTable.has(localId), 0x254 /* "unrecognized id in redirect table" */);
+            this.redirectTable.set(localId, storageId);
+            // set identity (id -> id) entry
+            this.redirectTable.set(storageId, storageId);
+        }
+    }
+    getPendingBlobs() {
+        const blobs = {};
+        for (const [key, entry] of this.pendingBlobs) {
+            blobs[key] = { blob: (0, common_utils_1.bufferToString)(entry.blob, "base64") };
         }
-        (0, common_utils_1.assert)(this.detachedBlobIds.size === 0, 0x255 /* "detached blob id absent in redirect table" */);
-        this.redirectTable = table;
+        return blobs;
     }
 }
 exports.BlobManager = BlobManager;