@fluidframework/container-runtime 2.0.0-dev.4.4.0.162574 → 2.0.0-dev.5.3.2.178189

package/src/summary/summarizer.ts

@@ -5,17 +5,17 @@
 
 import { EventEmitter } from "events";
 import { Deferred } from "@fluidframework/common-utils";
-import { ITelemetryLogger } from "@fluidframework/common-definitions";
-import { ILoader, LoaderHeader } from "@fluidframework/container-definitions";
-import { UsageError } from "@fluidframework/container-utils";
-import { DriverHeader } from "@fluidframework/driver-definitions";
-import { requestFluidObject } from "@fluidframework/runtime-utils";
 import {
+	ITelemetryLoggerExt,
 	ChildLogger,
 	IFluidErrorBase,
 	LoggingError,
 	wrapErrorAndLog,
 } from "@fluidframework/telemetry-utils";
+import { ILoader, LoaderHeader } from "@fluidframework/container-definitions";
+import { UsageError } from "@fluidframework/container-utils";
+import { DriverHeader } from "@fluidframework/driver-definitions";
+import { requestFluidObject } from "@fluidframework/runtime-utils";
 import { FluidObject, IFluidHandleContext, IRequest } from "@fluidframework/core-interfaces";
 import { ISummaryConfiguration } from "../containerRuntime";
 import { ICancellableSummarizerController } from "./runWhileConnectedCoordinator";
@@ -47,7 +47,7 @@ export class SummarizingWarning
 		super(errorMessage);
 	}
 
-	static wrap(error: any, logged: boolean = false, logger: ITelemetryLogger) {
+	static wrap(error: any, logged: boolean = false, logger: ITelemetryLoggerExt) {
 		const newErrorFn = (errMsg: string) => new SummarizingWarning(errMsg, logged);
 		return wrapErrorAndLog<SummarizingWarning>(error, newErrorFn, logger);
 	}
@@ -66,7 +66,7 @@ export class Summarizer extends EventEmitter implements ISummarizer {
 		return this;
 	}
 
-	private readonly logger: ITelemetryLogger;
+	private readonly logger: ITelemetryLoggerExt;
 	private runningSummarizer?: RunningSummarizer;
 	private _disposed: boolean = false;
 	private starting: boolean = false;
@@ -152,7 +152,7 @@ export class Summarizer extends EventEmitter implements ISummarizer {
 		// This will result in "summarizerClientDisconnected" stop reason recorded in telemetry,
 		// unless stop() was called earlier
 		this.dispose();
-		(this.runtime.disposeFn ?? this.runtime.closeFn)();
+		this.runtime.disposeFn();
 	}
 
 	private async runCore(onBehalfOf: string): Promise<SummarizerStopReason> {

package/src/summary/summarizerClientElection.ts

@@ -3,7 +3,8 @@
  * Licensed under the MIT License.
  */
 
-import { IEvent, IEventProvider, ITelemetryLogger } from "@fluidframework/common-definitions";
+import { IEvent, IEventProvider } from "@fluidframework/common-definitions";
+import { ITelemetryLoggerExt } from "@fluidframework/telemetry-utils";
 import { TypedEventEmitter } from "@fluidframework/common-utils";
 import { IClientDetails, MessageType } from "@fluidframework/protocol-definitions";
 import {
@@ -55,7 +56,7 @@ export class SummarizerClientElection
 	}
 
 	constructor(
-		private readonly logger: ITelemetryLogger,
+		private readonly logger: ITelemetryLoggerExt,
 		private readonly summaryCollection: IEventProvider<ISummaryCollectionOpEvents>,
 		public readonly clientElection: IOrderedClientElection,
 		private readonly maxOpsSinceLastSummary: number,

package/src/summary/summarizerHeuristics.ts

@@ -3,7 +3,7 @@
  * Licensed under the MIT License.
  */
 
-import { ITelemetryLogger } from "@fluidframework/common-definitions";
+import { ITelemetryLoggerExt } from "@fluidframework/telemetry-utils";
 import { Timer } from "@fluidframework/common-utils";
 import { ISummaryConfigurationHeuristics } from "../containerRuntime";
 import {
@@ -103,7 +103,7 @@ export class SummarizeHeuristicRunner implements ISummarizeHeuristicRunner {
 		private readonly heuristicData: ISummarizeHeuristicData,
 		private readonly configuration: ISummaryConfigurationHeuristics,
 		trySummarize: (reason: SummarizeReason) => void,
-		private readonly logger: ITelemetryLogger,
+		private readonly logger: ITelemetryLoggerExt,
 		private readonly summarizeStrategies: ISummaryHeuristicStrategy[] = getDefaultSummaryHeuristicStrategies(),
 	) {
 		this.idleTimer = new Timer(this.idleTime, () => this.runSummarize("idle"));

package/src/summary/summarizerNode/index.ts

@@ -7,6 +7,7 @@ export {
 	IFetchSnapshotResult,
 	ISummarizerNodeRootContract,
 	RefreshSummaryResult,
+	ValidateSummaryResult,
 } from "./summarizerNodeUtils";
 export { IRootSummarizerNode, createRootSummarizerNode } from "./summarizerNode";
 export { IRootSummarizerNodeWithGC, createRootSummarizerNodeWithGC } from "./summarizerNodeWithGc";

package/src/summary/summarizerNode/summarizerNode.ts

@@ -3,6 +3,7 @@
  * Licensed under the MIT License.
  */
 
+import { ITelemetryErrorEvent } from "@fluidframework/core-interfaces";
 import {
 	ISummarizerNode,
 	ISummarizerNodeConfig,
@@ -20,7 +21,13 @@ import {
 	ISnapshotTree,
 	SummaryObject,
 } from "@fluidframework/protocol-definitions";
-import { ITelemetryErrorEvent, ITelemetryLogger } from "@fluidframework/common-definitions";
+import {
+	ITelemetryLoggerExt,
+	ChildLogger,
+	LoggingError,
+	PerformanceEvent,
+	TelemetryDataTag,
+} from "@fluidframework/telemetry-utils";
 import { assert, unreachableCase } from "@fluidframework/common-utils";
 import {
 	convertToSummaryTree,
@@ -28,12 +35,6 @@ import {
 	mergeStats,
 	ReadAndParseBlob,
 } from "@fluidframework/runtime-utils";
-import {
-	ChildLogger,
-	LoggingError,
-	PerformanceEvent,
-	TelemetryDataTag,
-} from "@fluidframework/telemetry-utils";
 import {
 	EscapedPath,
 	ICreateChildDetails,
@@ -44,6 +45,7 @@ import {
 	parseSummaryTreeForSubtrees,
 	RefreshSummaryResult,
 	SummaryNode,
+	ValidateSummaryResult,
 } from "./summarizerNodeUtils";
 
 export interface IRootSummarizerNode extends ISummarizerNode, ISummarizerNodeRootContract {}
@@ -76,21 +78,21 @@ export class SummarizerNode implements IRootSummarizerNode {
 	private wipLocalPaths: { localPath: EscapedPath; additionalPath?: EscapedPath } | undefined;
 	private wipSkipRecursion = false;
 
-	protected readonly logger: ITelemetryLogger;
+	protected readonly logger: ITelemetryLoggerExt;
 
 	/**
 	 * Do not call constructor directly.
 	 * Use createRootSummarizerNode to create root node, or createChild to create child nodes.
 	 */
 	public constructor(
-		baseLogger: ITelemetryLogger,
+		baseLogger: ITelemetryLoggerExt,
 		private readonly summarizeInternalFn: SummarizeInternalFn,
 		config: ISummarizerNodeConfig,
 		private _changeSequenceNumber: number,
 		/** Undefined means created without summary */
 		private _latestSummary?: SummaryNode,
 		private readonly initialSummary?: IInitialSummary,
-		protected wipSummaryLogger?: ITelemetryLogger,
+		protected wipSummaryLogger?: ITelemetryLoggerExt,
 		/** A unique id of this node to be logged when sending telemetry. */
 		protected telemetryNodeId?: string,
 	) {
@@ -106,7 +108,7 @@ export class SummarizerNode implements IRootSummarizerNode {
 		});
 	}
 
-	public startSummary(referenceSequenceNumber: number, summaryLogger: ITelemetryLogger) {
+	public startSummary(referenceSequenceNumber: number, summaryLogger: ITelemetryLoggerExt) {
 		assert(
 			this.wipSummaryLogger === undefined,
 			0x19f /* "wipSummaryLogger should not be set yet in startSummary" */,
@@ -192,24 +194,117 @@ export class SummarizerNode implements IRootSummarizerNode {
 	}
 
 	/**
-	 * Complete the WIP summary for the given proposalHandle
+	 * Validates that the in-progress summary is correct, i.e., summarize should have run for all non-skipped
+	 * nodes. This will only be called for the root summarizer node and is called by it recursively on all child nodes.
+	 *
+	 * @returns ValidateSummaryResult which contains a boolean success indicating whether the validation was successful.
+	 * In case of failure, additional information is returned indicating type of failure and where it was.
+	 */
+	public validateSummary(): ValidateSummaryResult {
+		return this.validateSummaryCore(false /* parentSkipRecursion */);
+	}
+
+	/**
+	 * Validates that the in-progress summary is correct for all nodes, i.e., summarize should have run for all
+	 * non-skipped nodes.
+	 * @param parentSkipRecursion - true if the parent of this node skipped recursing the child nodes when summarizing.
+	 * In that case, the children will not have work-in-progress state.
+	 *
+	 * @returns ValidateSummaryResult which contains a boolean success indicating whether the validation was successful.
+	 * In case of failure, additional information is returned indicating type of failure and where it was.
+	 */
+	protected validateSummaryCore(parentSkipRecursion: boolean): ValidateSummaryResult {
+		if (this.wasSummarizeMissed(parentSkipRecursion)) {
+			return {
+				success: false,
+				reason: "NodeDidNotSummarize",
+				id: {
+					tag: TelemetryDataTag.CodeArtifact,
+					value: this.telemetryNodeId,
+				},
+				// These errors are usually transient and should go away when summarize is retried.
+				retryAfterSeconds: 1,
+			};
+		}
+		if (parentSkipRecursion) {
+			return { success: true };
+		}
+
+		for (const child of this.children.values()) {
+			const result = child.validateSummaryCore(this.wipSkipRecursion || parentSkipRecursion);
+			// If any child fails, return the failure.
+			if (!result.success) {
+				return result;
+			}
+		}
+		return { success: true };
+	}
+
+	private wasSummarizeMissed(parentSkipRecursion: boolean): boolean {
+		assert(
+			this.wipSummaryLogger !== undefined,
+			0x6fc /* wipSummaryLogger should have been set in startSummary or ctor */,
+		);
+		assert(this.wipReferenceSequenceNumber !== undefined, 0x6fd /* Not tracking a summary */);
+
+		// If the parent node skipped recursion, it did not call summarize on this node. So, summarize was not missed
+		// but was intentionally not called.
+		// Otherwise, summarize should have been called on this node and wipLocalPaths must be set.
+		if (parentSkipRecursion || this.wipLocalPaths !== undefined) {
+			return false;
+		}
+
+		/**
+		 * The absence of wip local path indicates that summarize was not called for this node. Return failure.
+		 * This can happen if:
+		 * 1. A child node was created after summarize was already called on the parent. For example, a data store
+		 * is realized (loaded) after summarize was called on it creating summarizer nodes for its DDSes. In this case,
+		 * parentSkipRecursion will be true and the if block above would handle it.
+		 * 2. A new node was created but summarize was never called on it. This can mean that the summary that is
+		 * generated may not have the data from this node. We should not continue, log and throw an error. This
+		 * will help us identify these cases and take appropriate action.
+		 *
+		 * This happens due to scenarios such as data store created during summarize. Such errors should go away when
+		 * summarize is attempted again.
+		 */
+		return true;
+	}
+
+	/**
+	 * Called after summary has been uploaded to the server. Add the work-in-progress state to the pending summary
+	 * queue. We track this until we get an ack from the server for this summary.
+	 * @param proposalHandle - The handle of the summary that was uploaded to the server.
 	 */
-	public completeSummary(proposalHandle: string) {
-		this.completeSummaryCore(proposalHandle, undefined, false);
+	public completeSummary(proposalHandle: string, validate: boolean) {
+		this.completeSummaryCore(
+			proposalHandle,
+			undefined /* parentPath */,
+			false /* parentSkipRecursion */,
+			validate,
+		);
 	}
 
 	/**
-	 * Recursive implementation for completeSummary, with additional internal-only parameters
+	 * Recursive implementation for completeSummary, with additional internal-only parameters.
+	 * @param proposalHandle - The handle of the summary that was uploaded to the server.
+	 * @param parentPath - The path of the parent node which is used to build the path of this node.
+	 * @param parentSkipRecursion - true if the parent of this node skipped recursing the child nodes when summarizing.
+	 * In that case, the children will not have work-in-progress state.
+	 * @param validate - true to validate that the in-progress summary is correct for all nodes.
 	 */
 	protected completeSummaryCore(
 		proposalHandle: string,
 		parentPath: EscapedPath | undefined,
 		parentSkipRecursion: boolean,
+		validate: boolean,
 	) {
-		assert(
-			this.wipSummaryLogger !== undefined,
-			0x1a3 /* "wipSummaryLogger should have been set in startSummary or ctor" */,
-		);
+		if (validate && this.wasSummarizeMissed(parentSkipRecursion)) {
+			this.throwUnexpectedError({
+				eventName: "NodeDidNotSummarize",
+				proposalHandle,
+			});
+		}
+
 		assert(this.wipReferenceSequenceNumber !== undefined, 0x1a4 /* "Not tracking a summary" */);
 		let localPathsToUse = this.wipLocalPaths;
 
@@ -238,22 +333,9 @@ export class SummarizerNode implements IRootSummarizerNode {
 			}
 		}
 
-		/**
-		 * The absence of wip local path indicates that summarize was not called for this node. This can happen if:
-		 * 1. A child node was created after summarize was already called on the parent. For example, a data store
-		 * is realized (loaded) after summarize was called on it creating summarizer nodes for its DDSes. In this case,
-		 * parentSkipRecursion will be true and the if block above would handle it.
-		 * 2. A new node was created but summarize was never called on it. This can mean that the summary that is
-		 * generated may not have the data from this node. We should not continue, log and throw an error. This
-		 * will help us identify these cases and take appropriate action.
-		 */
-		if (localPathsToUse === undefined) {
-			this.throwUnexpectedError({
-				eventName: "NodeNotSummarized",
-				proposalHandle,
-			});
-		}
-
+		// If localPathsToUse is undefined, it means summarize didn't run for this node and in that case the validate
+		// step should have failed.
+		assert(localPathsToUse !== undefined, 0x6fe /* summarize didn't run for node */);
 		const summary = new SummaryNode({
 			...localPathsToUse,
 			referenceSequenceNumber: this.wipReferenceSequenceNumber,
@@ -265,6 +347,7 @@ export class SummarizerNode implements IRootSummarizerNode {
 				proposalHandle,
 				fullPathForChildren,
 				this.wipSkipRecursion || parentSkipRecursion,
+				validate,
 			);
 		}
 		// Note that this overwrites existing pending summary with
@@ -306,7 +389,7 @@ export class SummarizerNode implements IRootSummarizerNode {
 		summaryRefSeq: number,
 		fetchLatestSnapshot: () => Promise<IFetchSnapshotResult>,
 		readAndParseBlob: ReadAndParseBlob,
-		correlatedSummaryLogger: ITelemetryLogger,
+		correlatedSummaryLogger: ITelemetryLoggerExt,
 	): Promise<RefreshSummaryResult> {
 		const eventProps: {
 			proposalHandle: string | undefined;
@@ -449,7 +532,7 @@ export class SummarizerNode implements IRootSummarizerNode {
 		snapshotTree: ISnapshotTree,
 		basePath: EscapedPath | undefined,
 		localPath: EscapedPath,
-		correlatedSummaryLogger: ITelemetryLogger,
+		correlatedSummaryLogger: ITelemetryLoggerExt,
 		readAndParseBlob: ReadAndParseBlob,
 	): Promise<void> {
 		// Possible re-entrancy. If we have already seen a summary later than this one, ignore it.
@@ -746,7 +829,7 @@ export class SummarizerNode implements IRootSummarizerNode {
  * @param config - Configure behavior of summarizer node
  */
 export const createRootSummarizerNode = (
-	logger: ITelemetryLogger,
+	logger: ITelemetryLoggerExt,
 	summarizeInternalFn: SummarizeInternalFn,
 	changeSequenceNumber: number,
 	referenceSequenceNumber: number | undefined,

package/src/summary/summarizerNode/summarizerNodeUtils.ts

@@ -3,7 +3,7 @@
  * Licensed under the MIT License.
  */
 
-import { ITelemetryLogger } from "@fluidframework/common-definitions";
+import { ITelemetryLoggerExt, TelemetryDataTag } from "@fluidframework/telemetry-utils";
 import { ISnapshotTree, ISummaryTree, SummaryObject } from "@fluidframework/protocol-definitions";
 import { channelsTreeName, ISummaryTreeWithStats } from "@fluidframework/runtime-definitions";
 import { ReadAndParseBlob } from "@fluidframework/runtime-utils";
@@ -42,16 +42,39 @@ export interface IFetchSnapshotResult {
 	snapshotRefSeq: number;
 }
 
+/**
+ * Return type of validateSummary function. In case of success, the object returned should have success: true.
+ * In case of failure, the object returned should have success: false and additional properties to indicate what
+ * the failure was, where it was, can it be retried, etc.
+ */
+export type ValidateSummaryResult =
+	| {
+			success: true;
+	  }
+	| {
+			success: false;
+			/** The failure reason */
+			reason: string;
+			/** id of the node that failed during validation */
+			id: {
+				tag: TelemetryDataTag.CodeArtifact;
+				value: string | undefined;
+			};
+			/** If the error can be retried, time to wait before retrying */
+			retryAfterSeconds?: number;
+	  };
+
 export interface ISummarizerNodeRootContract {
-	startSummary(referenceSequenceNumber: number, summaryLogger: ITelemetryLogger): void;
-	completeSummary(proposalHandle: string): void;
+	startSummary(referenceSequenceNumber: number, summaryLogger: ITelemetryLoggerExt): void;
+	validateSummary(): ValidateSummaryResult;
+	completeSummary(proposalHandle: string, validate: boolean): void;
 	clearSummary(): void;
 	refreshLatestSummary(
 		proposalHandle: string | undefined,
 		summaryRefSeq: number,
 		fetchLatestSnapshot: () => Promise<IFetchSnapshotResult>,
 		readAndParseBlob: ReadAndParseBlob,
-		correlatedSummaryLogger: ITelemetryLogger,
+		correlatedSummaryLogger: ITelemetryLoggerExt,
 	): Promise<RefreshSummaryResult>;
 }
 

package/src/summary/summarizerNode/summarizerNodeWithGc.ts

@@ -3,8 +3,13 @@
  * Licensed under the MIT License.
  */
 
-import { ITelemetryLogger } from "@fluidframework/common-definitions";
-import { assert, LazyPromise } from "@fluidframework/common-utils";
+import {
+	ITelemetryLoggerExt,
+	LoggingError,
+	TelemetryDataTag,
+} from "@fluidframework/telemetry-utils";
+import { assert } from "@fluidframework/common-utils";
+import { LazyPromise } from "@fluidframework/core-utils";
 import { ISnapshotTree } from "@fluidframework/protocol-definitions";
 import {
 	CreateChildSummarizerNodeParam,
@@ -19,7 +24,6 @@ import {
 	ITelemetryContext,
 	IExperimentalIncrementalSummaryContext,
 } from "@fluidframework/runtime-definitions";
-import { LoggingError, TelemetryDataTag } from "@fluidframework/telemetry-utils";
 import { ReadAndParseBlob, unpackChildNodesUsedRoutes } from "@fluidframework/runtime-utils";
 import {
 	cloneGCData,
@@ -35,6 +39,7 @@ import {
 	ISummarizerNodeRootContract,
 	parseSummaryForSubtrees,
 	SummaryNode,
+	ValidateSummaryResult,
 } from "./summarizerNodeUtils";
 
 export interface IRootSummarizerNodeWithGC
@@ -44,7 +49,7 @@ export interface IRootSummarizerNodeWithGC
 // Extend SummaryNode to add used routes tracking to it.
 class SummaryNodeWithGC extends SummaryNode {
 	constructor(
-		public readonly serializedUsedRoutes: string,
+		public readonly serializedUsedRoutes: string | undefined,
 		summary: {
 			readonly referenceSequenceNumber: number;
 			readonly basePath: EscapedPath | undefined;
@@ -102,7 +107,7 @@ export class SummarizerNodeWithGC extends SummarizerNode implements IRootSummari
 	 * Use createRootSummarizerNodeWithGC to create root node, or createChild to create child nodes.
 	 */
 	public constructor(
-		logger: ITelemetryLogger,
+		logger: ITelemetryLoggerExt,
 		private readonly summarizeFn: (
 			fullTree: boolean,
 			trackState: boolean,
@@ -114,7 +119,7 @@ export class SummarizerNodeWithGC extends SummarizerNode implements IRootSummari
 		/** Undefined means created without summary */
 		latestSummary?: SummaryNode,
 		initialSummary?: IInitialSummary,
-		wipSummaryLogger?: ITelemetryLogger,
+		wipSummaryLogger?: ITelemetryLoggerExt,
 		private readonly getGCDataFn?: (fullGC?: boolean) => Promise<IGarbageCollectionData>,
 		getBaseGCDetailsFn?: () => Promise<IGarbageCollectionDetailsBase>,
 		/** A unique id of this node to be logged when sending telemetry. */
@@ -241,7 +246,7 @@ export class SummarizerNodeWithGC extends SummarizerNode implements IRootSummari
 	/**
 	 * Called during the start of a summary. Updates the work-in-progress used routes.
 	 */
-	public startSummary(referenceSequenceNumber: number, summaryLogger: ITelemetryLogger) {
+	public startSummary(referenceSequenceNumber: number, summaryLogger: ITelemetryLoggerExt) {
 		// If GC is disabled, skip setting wip used routes since we should not track GC state.
 		if (!this.gcDisabled) {
 			assert(
@@ -252,45 +257,86 @@ export class SummarizerNodeWithGC extends SummarizerNode implements IRootSummari
 		super.startSummary(referenceSequenceNumber, summaryLogger);
 	}
 
+	/**
+	 * Validates that the in-progress summary is correct for all nodes, i.e., GC should have run for non-skipped nodes.
+	 * @param parentSkipRecursion - true if the parent of this node skipped recursing the child nodes when running GC.
+	 * In that case, the children will not have work-in-progress state.
+	 *
+	 * @returns ValidateSummaryResult which contains a boolean success indicating whether the validation was successful.
+	 * In case of failure, additional information is returned indicating type of failure and where it was.
+	 */
+	protected validateSummaryCore(parentSkipRecursion: boolean): ValidateSummaryResult {
+		if (this.wasGCMissed()) {
+			return {
+				success: false,
+				reason: "NodeDidNotRunGC",
+				id: {
+					tag: TelemetryDataTag.CodeArtifact,
+					value: this.telemetryNodeId,
+				},
+				// These errors are usually transient and should go away when summarize / GC is retried.
+				retryAfterSeconds: 1,
+			};
+		}
+		return super.validateSummaryCore(parentSkipRecursion);
+	}
+
+	private wasGCMissed(): boolean {
+		// If GC is disabled, it should not have run so it was not missed.
+		// Otherwise, GC should have been called on this node and wipSerializedUsedRoutes must be set.
+		if (this.gcDisabled || this.wipSerializedUsedRoutes !== undefined) {
+			return false;
+		}
+		/**
+		 * The absence of wip used routes indicates that GC was not run on this node. This can happen if:
+		 * 1. A child node was created after GC was already run on the parent. For example, a data store
+		 * is realized (loaded) after GC was run on it creating summarizer nodes for its DDSes. In this
+		 * case, the parent will pass on used routes to the child nodes and it will have wip used routes.
+		 * 2. A new node was created but GC was never run on it. This can mean that the GC data generated
+		 * during summarize is incomplete.
+		 *
+		 * This happens due to scenarios such as data store created during summarize. Such errors should go away when
+		 * summarize is attempted again.
+		 */
+		return true;
+	}
+
 	/**
 	 * Called after summary has been uploaded to the server. Add the work-in-progress state to the pending
 	 * summary queue. We track this until we get an ack from the server for this summary.
+	 * @param proposalHandle - The handle of the summary that was uploaded to the server.
+	 * @param parentPath - The path of the parent node which is used to build the path of this node.
+	 * @param parentSkipRecursion - true if the parent of this node skipped recursing the child nodes when summarizing.
+	 * In that case, the children will not have work-in-progress state.
+	 * @param validate - true to validate that the in-progress summary is correct for all nodes.
 	 */
 	protected completeSummaryCore(
 		proposalHandle: string,
 		parentPath: EscapedPath | undefined,
 		parentSkipRecursion: boolean,
+		validate: boolean,
 	) {
+		if (validate && this.wasGCMissed()) {
+			this.throwUnexpectedError({
+				eventName: "NodeDidNotRunGC",
+				proposalHandle,
+			});
+		}
+
 		let wipSerializedUsedRoutes: string | undefined;
 		// If GC is disabled, don't set wip used routes.
 		if (!this.gcDisabled) {
 			wipSerializedUsedRoutes = this.wipSerializedUsedRoutes;
-			/**
-			 * The absence of wip used routes indicates that GC was not run on this node. This can happen if:
-			 * 1. A child node was created after GC was already run on the parent. For example, a data store
-			 * is realized (loaded) after GC was run on it creating summarizer nodes for its DDSes. In this
-			 * case, the used routes of the parent should be passed on the child nodes and it should be fine.
-			 * 2. A new node was created but GC was never run on it. This can mean that the GC data generated
-			 * during summarize is complete . We should not continue, log and throw an error. This will help us
-			 * identify these cases and take appropriate action.
-			 */
-			if (wipSerializedUsedRoutes === undefined) {
-				this.throwUnexpectedError({
-					eventName: "NodeDidNotRunGC",
-					proposalHandle,
-				});
-			}
 		}
 
-		super.completeSummaryCore(proposalHandle, parentPath, parentSkipRecursion);
+		super.completeSummaryCore(proposalHandle, parentPath, parentSkipRecursion, validate);
 
 		// If GC is disabled, skip setting pending summary with GC state.
 		if (!this.gcDisabled) {
 			const summaryNode = this.pendingSummaries.get(proposalHandle);
 			if (summaryNode !== undefined) {
 				const summaryNodeWithGC = new SummaryNodeWithGC(
-					// eslint-disable-next-line @typescript-eslint/no-non-null-assertion
-					wipSerializedUsedRoutes!,
+					wipSerializedUsedRoutes,
 					summaryNode,
 				);
 				this.pendingSummaries.set(proposalHandle, summaryNodeWithGC);
@@ -352,7 +398,7 @@ export class SummarizerNodeWithGC extends SummarizerNode implements IRootSummari
 		snapshotTree: ISnapshotTree,
 		basePath: EscapedPath | undefined,
 		localPath: EscapedPath,
-		correlatedSummaryLogger: ITelemetryLogger,
+		correlatedSummaryLogger: ITelemetryLoggerExt,
 		readAndParseBlob: ReadAndParseBlob,
 	): Promise<void> {
 		await this.refreshGCStateFromSnapshot(
@@ -620,7 +666,7 @@ export class SummarizerNodeWithGC extends SummarizerNode implements IRootSummari
  * @param baseGCDetailsP - Function to get the initial GC details of this node
  */
 export const createRootSummarizerNodeWithGC = (
-	logger: ITelemetryLogger,
+	logger: ITelemetryLoggerExt,
 	summarizeInternalFn: SummarizeInternalFn,
 	changeSequenceNumber: number,
 	referenceSequenceNumber: number | undefined,

package/src/summary/summarizerTypes.ts

@@ -3,13 +3,9 @@
  * Licensed under the MIT License.
  */
 
-import {
-	IEvent,
-	IEventProvider,
-	ITelemetryLogger,
-	ITelemetryProperties,
-} from "@fluidframework/common-definitions";
-import { ITelemetryLoggerPropertyBag } from "@fluidframework/telemetry-utils";
+import { IEvent, IEventProvider } from "@fluidframework/common-definitions";
+import { ITelemetryProperties } from "@fluidframework/core-interfaces";
+import { ITelemetryLoggerExt, ITelemetryLoggerPropertyBag } from "@fluidframework/telemetry-utils";
 import { ContainerWarning, IDeltaManager } from "@fluidframework/container-definitions";
 import {
 	ISequencedDocumentMessage,
@@ -72,17 +68,17 @@ export interface IConnectableRuntime {
 }
 
 export interface ISummarizerRuntime extends IConnectableRuntime {
-	readonly logger: ITelemetryLogger;
+	readonly logger: ITelemetryLoggerExt;
 	/** clientId of parent (non-summarizing) container that owns summarizer container */
 	readonly summarizerClientId: string | undefined;
 	readonly deltaManager: IDeltaManager<ISequencedDocumentMessage, IDocumentMessage>;
-	disposeFn?(): void;
+	disposeFn(): void;
 	closeFn(): void;
-	on?(
+	on(
 		event: "op",
 		listener: (op: ISequencedDocumentMessage, runtimeMessage?: boolean) => void,
 	): this;
-	off?(
+	off(
 		event: "op",
 		listener: (op: ISequencedDocumentMessage, runtimeMessage?: boolean) => void,
 	): this;
@@ -107,12 +103,12 @@ export interface IRefreshSummaryAckOptions {
 	/** Reference sequence number from the ack's summary op */
 	readonly summaryRefSeq: number;
 	/** Telemetry logger to which telemetry events will be forwarded. */
-	readonly summaryLogger: ITelemetryLogger;
+	readonly summaryLogger: ITelemetryLoggerExt;
 }
 
 export interface ISubmitSummaryOptions extends ISummarizeOptions {
 	/** Logger to use for correlated summary events */
-	readonly summaryLogger: ITelemetryLogger;
+	readonly summaryLogger: ITelemetryLoggerExt;
 	/** Tells when summary process should be cancelled */
 	readonly cancellationToken: ISummaryCancellationToken;
 }
@@ -218,6 +214,13 @@ export type SubmitSummaryResult =
 	| IUploadSummaryResult
 	| ISubmitSummaryOpResult;
 
+/** The stages of Summarize, used to describe how far progress succeeded in case of a failure at a later stage. */
+export type SummaryStage = SubmitSummaryResult["stage"] | "unknown";
+/** The data in summarizer result when submit summary stage fails. */
+export interface SubmitSummaryFailureData {
+	stage: SummaryStage;
+}
+
 export interface IBroadcastSummaryResult {
 	readonly summarizeOp: ISummaryOpMessage;
 	readonly broadcastDuration: number;
@@ -248,7 +251,9 @@ export type SummarizeResultPart<TSuccess, TFailure = undefined> =
 
 export interface ISummarizeResults {
 	/** Resolves when we generate, upload, and submit the summary. */
-	readonly summarySubmitted: Promise<SummarizeResultPart<SubmitSummaryResult>>;
+	readonly summarySubmitted: Promise<
+		SummarizeResultPart<SubmitSummaryResult, SubmitSummaryFailureData>
+	>;
 	/** Resolves when we observe our summarize op broadcast. */
 	readonly summaryOpBroadcasted: Promise<SummarizeResultPart<IBroadcastSummaryResult>>;
 	/** Resolves when we receive a summaryAck or summaryNack. */