@aztec/p2p 0.0.1-commit.29c6b1a3 → 0.0.1-commit.2c0ee1788

package/src/services/reqresp/batch-tx-requester/README.md

@@ -1,14 +1,14 @@
 # BatchTxRequester
 
-The `BatchTxRequester` is a specialized P2P service that aggressively fetches missing transactions from peers when a node receives a block proposal but lacks some of the referenced transactions. This is critical for validators who need all transactions to attest to a proposal.
+The `BatchTxRequester` is a specialized P2P service that aggressively fetches missing transactions from peers when a node lacks some of the referenced transactions. It serves two pathways: (1) block proposals, where validators need all transactions to attest, and (2) block proving, where provers need all transactions to generate proofs.
 
 ## Overview
 
-When a validator receives a block proposal, they must verify all transactions in the block. If some transactions are missing from the local mempool (e.g., due to gossip delays), the `BatchTxRequester` kicks in to fetch them via direct peer-to-peer requests before the attestation deadline.
+When a validator receives a block proposal or a prover needs to prove a block, they must have all transactions. If some transactions are missing from the local mempool (e.g., due to gossip delays), the `BatchTxRequester` kicks in to fetch them via direct peer-to-peer requests before the deadline.
 
 ```
 ┌─────────────────────────────────────────────────────────────────────────────┐
-│                           Block Proposal Received                           │
+│                      Block Proposal or Block Received                       │
 │                     (contains hashes of N transactions)                     │
 └─────────────────────────────────────────────────────────────────────────────┘
                                       │
@@ -71,7 +71,7 @@ The requester classifies peers into three categories to optimize fetching:
 ### Blind Phase → Smart Phase Transition
 
 Peers transition from "dumb" to "smart" when they respond with a valid `BlockTxsResponse` containing:
-1. A matching `blockHash`
+1. A matching `archiveRoot`
 2. A non-empty `txIndices` BitVector indicating which transactions they have
 3. At least one transaction we're still missing
 
@@ -81,10 +81,10 @@ Peers transition from "dumb" to "smart" when they respond with a valid `BlockTxs
 │  ┌────────────────────────────────────────────────────────────────────────┐  │
 │  │  Initial State: All peers are "dumb" (except pinned peer)              │  │
 │  │                                                                        │  │
-│  │  Request: [blockHash, txHashes (full list), txIndices (BitVector)]     │  │
+│  │  Request: [archiveRoot, txHashes (full list), txIndices (BitVector)]   │  │
 │  │           └─ Include full hashes because peer may not have proposal    │  │
 │  │                                                                        │  │
-│  │  Response: [blockHash, txs[], txIndices (what peer has)]               │  │
+│  │  Response: [archiveRoot, txs[], txIndices (what peer has)]             │  │
 │  │            └─ Tells us exactly which txs this peer can provide         │  │
 │  └────────────────────────────────────────────────────────────────────────┘  │
 └──────────────────────────────────────────────────────────────────────────────┘
@@ -97,10 +97,10 @@ Peers transition from "dumb" to "smart" when they respond with a valid `BlockTxs
 │  ┌────────────────────────────────────────────────────────────────────────┐  │
 │  │  Peer promoted to "smart" - we know exactly what they have             │  │
 │  │                                                                        │  │
-│  │  Request: [blockHash, txIndices (BitVector only)]                      │  │
+│  │  Request: [archiveRoot, txIndices (BitVector only)]                    │  │
 │  │           └─ No need for full hashes, peer has the proposal            │  │
 │  │                                                                        │  │
-│  │  Response: [blockHash, txs[], txIndices (updated availability)]        │  │
+│  │  Response: [archiveRoot, txs[], txIndices (updated availability)]      │  │
 │  │            └─ May have received more txs since last response           │  │
 │  └────────────────────────────────────────────────────────────────────────┘  │
 └──────────────────────────────────────────────────────────────────────────────┘
@@ -152,7 +152,7 @@ The `BatchTxRequester` runs three types of workers concurrently:
 
 ```typescript
 class BlockTxsRequest {
-  blockHash: Fr;           // 32-byte hash of the proposed block header
+  archiveRoot: Fr;         // Archive root after the proposed block is applied
   txHashes: TxHashArray;   // Full tx hashes (for dumb peers without proposal)
   txIndices: BitVector;    // Which txs from proposal we're requesting (1 = want)
 }
@@ -162,7 +162,7 @@ class BlockTxsRequest {
 
 ```typescript
 class BlockTxsResponse {
-  blockHash: Fr;           // Echo back the block hash
+  archiveRoot: Fr;         // Echo back the proposal archive root
   txs: TxArray;            // Actual transaction data
   txIndices: BitVector;    // Which txs the peer has available (1 = have)
 }
@@ -170,6 +170,37 @@ class BlockTxsResponse {
 
 The `BitVector` is a compact representation where each bit corresponds to a transaction index in the block proposal. This allows efficient capability advertisement without repeating full hashes.
 
+## Cancellation
+
+All cancellation is managed by a single `RequestTracker` instance, shared across the entire collection
+flow. The `RequestTracker` owns the deadline, tracks which txs are still missing, and exposes a
+`cancellationToken` promise that resolves when the request should stop (deadline hit, all txs fetched,
+or external `cancel()` call).
+
+Cancellation propagates from the deepest stack level upward:
+
+```
+RequestTracker.finish()
+  ├── resolves cancellationToken promise
+  │
+  ├── BatchTxRequester workers (deepest)
+  │     ├── shouldStop() checks requestTracker.cancelled → exit loop
+  │     ├── sleepClampedToDeadline races sleep vs cancellationToken → wakes
+  │     └── semaphore.acquire races vs cancellationToken → wakes
+  │           │
+  │           ▼ workers settle → txQueue.end() → generator returns
+  │
+  ├── Node collection loops
+  │     ├── notFinished() checks requestTracker.cancelled → exit loop
+  │     └── inter-retry sleep races vs cancellationToken → wakes
+  │           │
+  │           ▼ all node loops settle
+  │
+  └── collectFast (outermost)
+        awaits Promise.allSettled([reqresp, nodes]) → settles after inner tasks
+        finally: requestTracker.cancel() (idempotent), cleanup
+```
+
 ## Key Files
 
 | File | Description |
@@ -179,15 +210,16 @@ The `BitVector` is a compact representation where each bit corresponds to a tran
 | `peer_collection.ts` | Manages peer classification (dumb/smart/bad) and rate limiting |
 | `interface.ts` | Type definitions for dependencies |
 | `../protocols/block_txs/` | Wire protocol definitions (`BlockTxsRequest`, `BlockTxsResponse`, `BitVector`) |
+| `../../tx_collection/request_tracker.ts` | Centralized deadline, missing tx tracking, and cancellation signal |
 
 ## Stopping Conditions
 
-The `BatchTxRequester` stops when any of these conditions are met:
+The `BatchTxRequester` stops when any of these conditions are met, all managed by the `RequestTracker`:
 
-1. **All transactions fetched** - Success!
-2. **Deadline exceeded** - Timeout configured by caller
-3. **Abort signal** - External cancellation
-4. **No transactions to fetch** - Nothing was missing
+1. **All transactions fetched** - `markFetched()` removes the last missing tx, triggering `finish()`
+2. **Deadline exceeded** - `setTimeout` in `RequestTracker` fires, triggering `finish()`
+3. **External cancellation** - `RequestTracker.cancel()` called (e.g., from `stop()`, `stopCollectingForBlocksUpTo`)
+4. **No transactions to fetch** - Empty hash set at construction, `RequestTracker` finishes immediately
 
 ## Configuration
 
@@ -228,11 +260,15 @@ Request to peer fails
 ## Usage Example
 
 ```typescript
+const requestTracker = RequestTracker.create(
+  missingTxHashes,                          // TxHash[] - what we need
+  new Date(Date.now() + 5_000),             // deadline
+);
+
 const requester = new BatchTxRequester(
-  missingTxHashes,      // TxHash[] - what we need
-  blockProposal,        // BlockProposal - the proposal we're attesting to
-  pinnedPeer,           // PeerId | undefined - who sent us the proposal
-  timeoutMs,            // number - how long to try
+  requestTracker,       // IRequestTracker - tracks missing txs, deadline, and cancellation
+  blockTxsSource,       // BlockTxsSource - the proposal or block we need txs for
+  pinnedPeer,           // PeerId | undefined - peer expected to have the txs
   p2pService,           // BatchTxRequesterLibP2PService
 );
 
@@ -268,19 +304,22 @@ const txs = await BatchTxRequester.collectAllTxs(requester.run());
 ┌───────────────────────────────────┐   ┌─────────────────────────────────────┐
 │        FastTxCollection           │   │         SlowTxCollection            │
 │                                   │   │                                     │
-│  Time-critical: attestations      │   │  Background: unproven blocks        │
+│  Time-critical: proposals/proving │   │  Background: unproven blocks        │
 │                                   │   │                                     │
 │  1. Try RPC nodes first (fast)    │   │  Periodic polling of RPC nodes      │
 │  2. Fall back to BatchTxRequester │   │  and peers for missing txs          │
 │                                   │   │                                     │
+│  Creates RequestTracker per       │   │                                     │
+│  request with deadline            │   │                                     │
 └───────────────────┬───────────────┘   └─────────────────────────────────────┘
                     │
-                    │ For 'proposal' requests
+                    │ For 'proposal' and 'block' requests
                     ▼
 ┌─────────────────────────────────────────────────────────────────────────────┐
 │                           BatchTxRequester                                  │
 │                                                                             │
 │  Aggressive parallel fetching from multiple peers                           │
+│  Shares RequestTracker with FastTxCollection for unified cancellation       │
 │  Uses BLOCK_TXS sub-protocol for efficient batching                         │
 └───────────────────┬─────────────────────────────────────────────────────────┘
                     │

package/src/services/reqresp/batch-tx-requester/batch_tx_requester.ts

@@ -1,17 +1,16 @@
 import { chunkWrapAround } from '@aztec/foundation/collection';
-import { TimeoutError } from '@aztec/foundation/error';
 import { type Logger, createLogger } from '@aztec/foundation/log';
 import { FifoMemoryQueue, type ISemaphore, Semaphore } from '@aztec/foundation/queue';
 import { sleep } from '@aztec/foundation/sleep';
-import { DateProvider, executeTimeout } from '@aztec/foundation/timer';
-import { type BlockProposal, PeerErrorSeverity } from '@aztec/stdlib/p2p';
+import { DateProvider } from '@aztec/foundation/timer';
+import { PeerErrorSeverity } from '@aztec/stdlib/p2p';
 import { Tx, TxArray, TxHash } from '@aztec/stdlib/tx';
 
 import type { PeerId } from '@libp2p/interface';
-import { peerIdFromString } from '@libp2p/peer-id';
 
+import type { IRequestTracker } from '../../tx_collection/request_tracker.js';
 import { ReqRespSubProtocol } from '.././interface.js';
-import { BlockTxsRequest, BlockTxsResponse } from '.././protocols/index.js';
+import { BlockTxsRequest, BlockTxsResponse, type BlockTxsSource } from '.././protocols/index.js';
 import { ReqRespStatus } from '.././status.js';
 import {
   DEFAULT_BATCH_TX_REQUESTER_BAD_PEER_THRESHOLD,
@@ -20,7 +19,7 @@ import {
   DEFAULT_BATCH_TX_REQUESTER_TX_BATCH_SIZE,
 } from './config.js';
 import type { BatchTxRequesterLibP2PService, BatchTxRequesterOptions, ITxMetadataCollection } from './interface.js';
-import { MissingTxMetadata, MissingTxMetadataCollection } from './missing_txs.js';
+import { MissingTxMetadataCollection } from './missing_txs.js';
 import { type IPeerCollection, PeerCollection } from './peer_collection.js';
 import { BatchRequestTxValidator, type IBatchRequestTxValidator } from './tx_validator.js';
 
@@ -42,16 +41,14 @@ import { BatchRequestTxValidator, type IBatchRequestTxValidator } from './tx_val
  *    - Is the peer which was unable to send us successful response N times in a row
  * */
 export class BatchTxRequester {
-  private readonly blockProposal: BlockProposal;
+  private readonly requestTracker: IRequestTracker;
+  private readonly blockTxsSource: BlockTxsSource;
   private readonly pinnedPeer: PeerId | undefined;
-  private readonly timeoutMs: number;
   private readonly p2pService: BatchTxRequesterLibP2PService;
   private readonly logger: Logger;
-  private readonly dateProvider: DateProvider;
   private readonly opts: BatchTxRequesterOptions;
   private readonly peers: IPeerCollection;
   private readonly txsMetadata: ITxMetadataCollection;
-  private readonly deadline: number;
   private readonly smartRequesterSemaphore: ISemaphore;
   private readonly txQueue: FifoMemoryQueue<Tx>;
   private readonly txValidator: IBatchRequestTxValidator;
@@ -60,21 +57,19 @@ export class BatchTxRequester {
   private readonly txBatchSize: number;
 
   constructor(
-    missingTxs: TxHash[],
-    blockProposal: BlockProposal,
+    requestTracker: IRequestTracker,
+    blockTxsSource: BlockTxsSource,
     pinnedPeer: PeerId | undefined,
-    timeoutMs: number,
     p2pService: BatchTxRequesterLibP2PService,
     logger?: Logger,
     dateProvider?: DateProvider,
     opts?: BatchTxRequesterOptions,
   ) {
-    this.blockProposal = blockProposal;
+    this.requestTracker = requestTracker;
+    this.blockTxsSource = blockTxsSource;
     this.pinnedPeer = pinnedPeer;
-    this.timeoutMs = timeoutMs;
     this.p2pService = p2pService;
     this.logger = logger ?? createLogger('p2p:reqresp_batch');
-    this.dateProvider = dateProvider ?? new DateProvider();
     this.opts = opts ?? {};
 
     this.smartParallelWorkerCount =
@@ -82,25 +77,22 @@ export class BatchTxRequester {
     this.dumbParallelWorkerCount =
       this.opts.dumbParallelWorkerCount ?? DEFAULT_BATCH_TX_REQUESTER_DUMB_PARALLEL_WORKER_COUNT;
     this.txBatchSize = this.opts.txBatchSize ?? DEFAULT_BATCH_TX_REQUESTER_TX_BATCH_SIZE;
-    this.deadline = this.dateProvider.now() + this.timeoutMs;
     this.txQueue = new FifoMemoryQueue(this.logger);
     this.txValidator = this.opts.txValidator ?? new BatchRequestTxValidator(this.p2pService.txValidatorConfig);
 
     if (this.opts.peerCollection) {
       this.peers = this.opts.peerCollection;
     } else {
-      const initialPeers = this.p2pService.connectionSampler.getPeerListSortedByConnectionCountAsc();
       const badPeerThreshold = this.opts.badPeerThreshold ?? DEFAULT_BATCH_TX_REQUESTER_BAD_PEER_THRESHOLD;
       this.peers = new PeerCollection(
-        initialPeers,
+        this.p2pService.connectionSampler,
         this.pinnedPeer,
-        this.dateProvider,
+        dateProvider ?? new DateProvider(),
         badPeerThreshold,
         this.p2pService.peerScoring,
       );
     }
-    const entries: Array<[string, MissingTxMetadata]> = missingTxs.map(h => [h.toString(), new MissingTxMetadata(h)]);
-    this.txsMetadata = new MissingTxMetadataCollection(entries, this.txBatchSize);
+    this.txsMetadata = new MissingTxMetadataCollection(requestTracker, this.txBatchSize);
     this.smartRequesterSemaphore = this.opts.semaphore ?? new Semaphore(0);
   }
 
@@ -108,40 +100,30 @@ export class BatchTxRequester {
    * Fetches all missing transactions and yields them  one by one
    * */
   public async *run(): AsyncGenerator<Tx, Tx | undefined, unknown> {
-    // Our timeout is represented in milliseconds but queue expects seconds
-    // We also want to make sure we wait at least 1 second in case of very low timeouts
-    const timeoutQueueAfter = Math.max(Math.ceil(this.timeoutMs / 1_000), 1);
     try {
       if (this.txsMetadata.getMissingTxHashes().size === 0) {
         return undefined;
       }
 
-      // Start workers in background
-      const workersPromise = executeTimeout(
-        () => Promise.allSettled([this.smartRequester(), this.dumbRequester(), this.pinnedPeerRequester()]),
-        this.timeoutMs,
-      ).finally(() => {
+      // Start workers in background. Workers stop themselves via requestTracker.checkCancelled().
+      const workersPromise = Promise.allSettled([
+        this.smartRequester(),
+        this.dumbRequester(),
+        this.pinnedPeerRequester(),
+      ]).finally(() => {
         this.txQueue.end();
       });
 
+      // Yield txs as workers put them on the queue. The queue's end() drains remaining items
+      // before returning null, so we don't lose any txs.
       while (true) {
-        const tx = await this.txQueue.get(timeoutQueueAfter);
+        const tx = await this.txQueue.get();
 
-        // null indicates that the queue has ended
         if (tx === null) {
           break;
         }
 
         yield tx;
-
-        if (this.shouldStop()) {
-          // Drain queue before ending
-          let remaining;
-          while ((remaining = this.txQueue.getImmediate()) !== undefined) {
-            yield remaining;
-          }
-          break;
-        }
       }
 
       this.unlockSmartRequesterSemaphores();
@@ -205,7 +187,7 @@ export class BatchTxRequester {
         return;
       }
 
-      const request = BlockTxsRequest.fromBlockProposalAndMissingTxs(this.blockProposal, txs);
+      const request = BlockTxsRequest.fromTxsSourceAndMissingTxs(this.blockTxsSource, txs);
       if (!request) {
         return;
       }
@@ -227,7 +209,6 @@ export class BatchTxRequester {
    * Starts dumb worker loops
    * */
   private async dumbRequester() {
-    const nextPeerIndex = this.makeRoundRobinIndexer();
     const nextBatchIndex = this.makeRoundRobinIndexer();
 
     // Chunk missing tx hashes into batches of txBatchSize, wrapping around to ensure no peer gets less than txBatchSize
@@ -249,8 +230,8 @@ export class BatchTxRequester {
       // If peer is dumb peer, we don't know yet if they received full blockProposal
       // there is solid chance that peer didn't receive proposal yet, thus we must send full hashes
       const includeFullHashesInRequestNotJustIndices = true;
-      const blockRequest = BlockTxsRequest.fromBlockProposalAndMissingTxs(
-        this.blockProposal,
+      const blockRequest = BlockTxsRequest.fromTxsSourceAndMissingTxs(
+        this.blockTxsSource,
         txs,
         includeFullHashesInRequestNotJustIndices,
       );
@@ -263,15 +244,9 @@ export class BatchTxRequester {
       return { blockRequest, txs };
     };
 
-    const nextPeer = () => {
-      const peers = this.peers.getDumbPeersToQuery();
-      const idx = nextPeerIndex(() => peers.length);
-      return idx === undefined ? undefined : peerIdFromString(peers[idx]);
-    };
-
-    const workerCount = Math.min(this.dumbParallelWorkerCount, this.peers.getAllPeers().size);
+    const workerCount = this.dumbParallelWorkerCount;
     const workers = Array.from({ length: workerCount }, (_, index) =>
-      this.dumbWorkerLoop(nextPeer, makeRequest, index + 1),
+      this.dumbWorkerLoop(this.peers.nextDumbPeerToQuery.bind(this.peers), makeRequest, index + 1),
     );
 
     await Promise.allSettled(workers);
@@ -332,17 +307,9 @@ export class BatchTxRequester {
    * Starts smart worker loops
    * */
   private async smartRequester() {
-    const nextPeerIndex = this.makeRoundRobinIndexer();
-
-    const nextPeer = () => {
-      const peers = this.peers.getSmartPeersToQuery();
-      const idx = nextPeerIndex(() => peers.length);
-      return idx === undefined ? undefined : peerIdFromString(peers[idx]);
-    };
-
     const makeRequest = (pid: PeerId) => {
       const txs = this.txsMetadata.getTxsToRequestFromThePeer(pid);
-      const blockRequest = BlockTxsRequest.fromBlockProposalAndMissingTxs(this.blockProposal, txs);
+      const blockRequest = BlockTxsRequest.fromTxsSourceAndMissingTxs(this.blockTxsSource, txs);
       if (!blockRequest) {
         return undefined;
       }
@@ -350,9 +317,8 @@ export class BatchTxRequester {
       return { blockRequest, txs };
     };
 
-    const workers = Array.from(
-      { length: Math.min(this.smartParallelWorkerCount, this.peers.getAllPeers().size) },
-      (_, index) => this.smartWorkerLoop(nextPeer, makeRequest, index + 1),
+    const workers = Array.from({ length: this.smartParallelWorkerCount }, (_, index) =>
+      this.smartWorkerLoop(this.peers.nextSmartPeerToQuery.bind(this.peers), makeRequest, index + 1),
     );
 
     await Promise.allSettled(workers);
@@ -378,7 +344,10 @@ export class BatchTxRequester {
   ) {
     try {
       this.logger.trace(`Smart worker ${workerIndex} started`);
-      await executeTimeout((_: AbortSignal) => this.smartRequesterSemaphore.acquire(), this.timeoutMs);
+      await Promise.race([this.smartRequesterSemaphore.acquire(), this.requestTracker.cancellationToken]);
+      if (this.requestTracker.checkCancelled()) {
+        return;
+      }
       this.logger.trace(`Smart worker ${workerIndex} acquired semaphore`);
 
       while (!this.shouldStop()) {
@@ -387,30 +356,25 @@ export class BatchTxRequester {
         if (weRanOutOfPeersToQuery) {
           this.logger.debug(`Worker loop smart: No more peers to query`);
 
-          // If there are no more dumb peers to query then none of our peers can become smart,
-          // thus we can simply exit this worker
-          const noMoreDumbPeersToQuery = this.peers.getDumbPeersToQuery().length === 0;
-          if (noMoreDumbPeersToQuery) {
-            // These might be either smart peers that will get unblocked after _some time_
-            const nextSmartPeerDelay = this.peers.getNextSmartPeerAvailabilityDelayMs();
-            const thereAreSomeRateLimitedSmartPeers = nextSmartPeerDelay !== undefined;
-            if (thereAreSomeRateLimitedSmartPeers) {
-              await this.sleepClampedToDeadline(nextSmartPeerDelay);
-              continue;
-            }
-
-            this.logger.debug(`Worker loop smart: No more smart peers to query killing ${workerIndex}`);
-            break;
+          // If we have rate limited peers wait for them.
+          const nextSmartPeerDelay = this.peers.getNextSmartPeerAvailabilityDelayMs();
+          const thereAreSomeRateLimitedSmartPeers = nextSmartPeerDelay !== undefined;
+          if (thereAreSomeRateLimitedSmartPeers) {
+            await this.sleepClampedToDeadline(nextSmartPeerDelay);
+            continue;
           }
 
-          // Otherwise there are still some dumb peers that could become smart.
           // We end up here when all known smart peers became temporarily unavailable via combination of
           // (bad, in-flight, or rate-limited) or in some weird scenario all current smart peers turn bad which is permanent
-          // but dumb peers still exist that could become smart.
+          // but there are dumb peers that could be promoted
+          // or new peer can join as dumb and be promoted later
           //
           // When a dumb peer responds with valid txIndices, it gets
           // promoted to smart and releases the semaphore, waking this worker.
-          await executeTimeout((_: AbortSignal) => this.smartRequesterSemaphore.acquire(), this.timeoutMs);
+          await Promise.race([this.smartRequesterSemaphore.acquire(), this.requestTracker.cancellationToken]);
+          if (this.requestTracker.checkCancelled()) {
+            break;
+          }
           this.logger.debug(`Worker loop smart: acquired next smart peer`);
           continue;
         }
@@ -437,11 +401,7 @@ export class BatchTxRequester {
         });
       }
     } catch (err: any) {
-      if (err instanceof TimeoutError) {
-        this.logger.debug(`Smart worker ${workerIndex} timed out waiting for semaphore`);
-      } else {
-        this.logger.error(`Smart worker ${workerIndex} encountered an error: ${err}`);
-      }
+      this.logger.error(`Smart worker ${workerIndex} encountered an error: ${err}`);
     } finally {
       this.logger.debug(`Smart worker ${workerIndex} finished`);
     }
@@ -489,9 +449,18 @@ export class BatchTxRequester {
    *   this implies we will query these peers couple of more times and give them a chance to "redeem" themselves before completely ignoring them
    */
   private handleFailResponseFromPeer(peerId: PeerId, responseStatus: ReqRespStatus) {
-    //TODO: Should we ban these peers?
     if (responseStatus === ReqRespStatus.FAILURE || responseStatus === ReqRespStatus.UNKNOWN) {
       this.peers.penalisePeer(peerId, PeerErrorSeverity.HighToleranceError);
+      this.peers.markPeerDumb(peerId);
+      this.txsMetadata.clearPeerData(peerId);
+      return;
+    }
+
+    // NOT_FOUND means the peer pruned its block proposal — it can no longer serve
+    // index-based requests, but this is a legitimate state so we don't penalize.
+    if (responseStatus === ReqRespStatus.NOT_FOUND) {
+      this.peers.markPeerDumb(peerId);
+      this.txsMetadata.clearPeerData(peerId);
       return;
     }
 
@@ -545,6 +514,9 @@ export class BatchTxRequester {
     });
 
     if (hasInvalidTx) {
+      this.logger.warn(`Penalizing peer ${peerId.toString()} for sending invalid transactions in batch response`, {
+        peerId,
+      });
       this.peers.penalisePeer(peerId, PeerErrorSeverity.LowToleranceError);
     } else {
       // If we have received successful response from the peer, they have "redeemed" themselves and not considered bad anymore
@@ -581,10 +553,9 @@ export class BatchTxRequester {
       return;
     }
 
-    // If block response is invalid we still want to query this peer in the future
-    // Because they sent successful response, so they might become smart peer in the future
-    // Or send us needed txs
-    if (!this.isBlockResponseValid(response)) {
+    const hasArchiveRootMismatch = this.blockTxsSource.archive.toString() !== response.archiveRoot.toString();
+    if (hasArchiveRootMismatch) {
+      this.handleArchiveRootMismatch(peerId, response);
       return;
     }
 
@@ -599,18 +570,28 @@ export class BatchTxRequester {
     this.markTxsPeerHas(peerId, response);
 
     // Unblock smart workers
-    if (this.peers.getSmartPeersToQuery().length <= this.smartParallelWorkerCount) {
-      this.smartRequesterSemaphore.release();
-    }
+    this.smartRequesterSemaphore.release();
   }
 
-  private isBlockResponseValid(response: BlockTxsResponse): boolean {
-    const blockIdsMatch = this.blockProposal.archive.toString() === response.blockHash.toString();
-    const peerHasSomeTxsFromProposal = !response.txIndices.isEmpty();
-    return blockIdsMatch && peerHasSomeTxsFromProposal;
+  /**
+   * Handles an archive root mismatch between local state and peer response.
+   *
+   * - Response archive is Fr.ZERO (peer pruned proposal, legitimate): marks peer dumb.
+   * - Non-zero archive mismatch (malicious response): penalises + marks dumb.
+   */
+  private handleArchiveRootMismatch(peerId: PeerId, response: BlockTxsResponse): void {
+    if (!response.archiveRoot.isZero()) {
+      this.peers.penalisePeer(peerId, PeerErrorSeverity.LowToleranceError);
+    }
+
+    this.peers.markPeerDumb(peerId);
+    this.txsMetadata.clearPeerData(peerId);
   }
 
   private peerHasSomeTxsWeAreMissing(_peerId: PeerId, response: BlockTxsResponse): boolean {
+    if (response.txIndices.isEmpty()) {
+      return false;
+    }
     const txsPeerHas = new Set(this.extractHashesPeerHasFromResponse(response).map(h => h.toString()));
     return this.txsMetadata.getMissingTxHashes().intersection(txsPeerHas).size > 0;
   }
@@ -624,7 +605,7 @@ export class BatchTxRequester {
   private extractHashesPeerHasFromResponse(response: BlockTxsResponse): Array<TxHash> {
     const hashes: TxHash[] = [];
     const indicesOfHashesPeerHas = new Set(response.txIndices.getTrueIndices());
-    this.blockProposal.txHashes.forEach((hash, idx) => {
+    this.blockTxsSource.txHashes.forEach((hash, idx) => {
       if (indicesOfHashesPeerHas.has(idx)) {
         hashes.push(hash);
       }
@@ -659,27 +640,14 @@ export class BatchTxRequester {
   }
 
   /*
-   * @returns true if all missing txs have been fetched */
-  private fetchedAllTxs() {
-    return Array.from(this.txsMetadata.values()).every(tx => tx.fetched);
-  }
-
-  /*
-   * Checks if the BatchTxRequester should stop fetching missing txs
-   * Conditions for stopping are:
-   * - There have been no missing transactions to start with
-   * - All transactions have been fetched
-   * - The deadline has been hit (no more time to fetch)
-   * - This process has been cancelled via abortSignal
-   *
-   * @returns true if BatchTxRequester should stop, otherwise false*/
+   * Checks if the BatchTxRequester should stop fetching missing txs.
+   * Delegates to requestTracker which covers: deadline hit, all txs fetched, or external cancellation. */
   private shouldStop() {
-    const aborted = this.opts.abortSignal?.aborted ?? false;
-    if (aborted) {
+    if (this.requestTracker.checkCancelled()) {
       this.unlockSmartRequesterSemaphores();
     }
 
-    return aborted || this.txsMetadata.size === 0 || this.fetchedAllTxs() || this.dateProvider.now() > this.deadline;
+    return this.requestTracker.checkCancelled();
   }
 
   /*
@@ -697,10 +665,9 @@ export class BatchTxRequester {
    * This ensures we don't sleep past the deadline.
    * */
   private async sleepClampedToDeadline(durationMs: number) {
-    const remaining = this.deadline - this.dateProvider.now();
-    const thereIsTimeRemaining = remaining > 0;
-    if (thereIsTimeRemaining) {
-      await sleep(Math.min(durationMs, remaining));
+    if (this.requestTracker.checkCancelled()) {
+      return;
     }
+    await Promise.race([sleep(durationMs), this.requestTracker.cancellationToken]);
   }
 }

package/src/services/reqresp/batch-tx-requester/interface.ts

@@ -6,7 +6,6 @@ import type { PeerId } from '@libp2p/interface';
 
 import type { ConnectionSampler } from '../connection-sampler/connection_sampler.js';
 import type { ReqRespInterface } from '../interface.js';
-import type { MissingTxMetadata } from './missing_txs.js';
 import type { IPeerCollection } from './peer_collection.js';
 import type { BatchRequestTxValidatorConfig, IBatchRequestTxValidator } from './tx_validator.js';
 
@@ -15,18 +14,17 @@ export interface IPeerPenalizer {
 }
 
 export interface ITxMetadataCollection {
-  size: number;
-  values(): IterableIterator<MissingTxMetadata>;
   getMissingTxHashes(): Set<string>;
+  markFetched(peerId: PeerId, tx: Tx): boolean;
   getTxsToRequestFromThePeer(peer: PeerId): TxHash[];
   markRequested(txHash: TxHash): void;
   markInFlightBySmartPeer(txHash: TxHash): void;
   markNotInFlightBySmartPeer(txHash: TxHash): void;
   alreadyFetched(txHash: TxHash): boolean;
   // Returns true if tx was marked as fetched, false if it was already marked as fetched
-  markFetched(peerId: PeerId, tx: Tx): boolean;
   markPeerHas(peerId: PeerId, txHashes: TxHash[]): void;
-  getFetchedTxs(): Tx[];
+  /** Remove all tx metadata associations for a peer (e.g. on demotion from smart to dumb). */
+  clearPeerData(peerId: PeerId): void;
 }
 
 /**
@@ -51,7 +49,6 @@ export interface BatchTxRequesterOptions {
   //Injectable for testing purposes
   semaphore?: ISemaphore;
   peerCollection?: IPeerCollection;
-  abortSignal?: AbortSignal;
   /** Optional tx validator for testing - if not provided, one is created from p2pService.txValidatorConfig */
   txValidator?: IBatchRequestTxValidator;
 }