Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

add retry logic on sync pipeline for rocksdb issue #6004

Merged
merged 13 commits into from
Oct 24, 2023
Merged
1 change: 1 addition & 0 deletions ethereum/eth/build.gradle
Original file line number Diff line number Diff line change
Expand Up @@ -58,6 +58,7 @@ dependencies {
implementation 'io.tmio:tuweni-bytes'
implementation 'io.tmio:tuweni-units'
implementation 'io.tmio:tuweni-rlp'
implementation 'org.rocksdb:rocksdbjni'

annotationProcessor "org.immutables:value"
implementation "org.immutables:value-annotations"
Expand Down
Original file line number Diff line number Diff line change
@@ -0,0 +1,64 @@
/*
* Copyright ConsenSys AG.
*
* Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
* the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
* an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
* specific language governing permissions and limitations under the License.
*
* SPDX-License-Identifier: Apache-2.0
*/
package org.hyperledger.besu.ethereum.eth.sync;

import org.hyperledger.besu.plugin.services.exception.StorageException;

matkt marked this conversation as resolved.
Show resolved Hide resolved
import java.util.EnumSet;
import java.util.Optional;

import org.rocksdb.RocksDBException;
import org.rocksdb.Status;

public final class StorageExceptionManager {

private static final EnumSet<Status.Code> RETRYABLE_STATUS_CODES =
EnumSet.of(Status.Code.TimedOut, Status.Code.TryAgain, Status.Code.Busy);

private static final long ERROR_THRESHOLD = 1000;

private static long retryableErrorCounter;
/**
* Determines if an operation can be retried based on the error received. This method checks if
* the cause of the StorageException is a RocksDBException. If it is, it retrieves the status code
* of the RocksDBException and checks if it is contained in the list of retryable {@link
* StorageExceptionManager.RETRYABLE_STATUS_CODES} status codes.
*
* @param e the StorageException to check
* @return true if the operation can be retried, false otherwise
*/
public static boolean canRetryOnError(final StorageException e) {
return Optional.of(e.getCause())
.filter(z -> z instanceof RocksDBException)
.map(RocksDBException.class::cast)
.map(RocksDBException::getStatus)
.map(Status::getCode)
.map(RETRYABLE_STATUS_CODES::contains)
.map(
result -> {
retryableErrorCounter++;
return result;
})
.orElse(false);
}

public static long getRetryableErrorCounter() {
return retryableErrorCounter;
}

public static boolean errorCountAtThreshold() {
return retryableErrorCounter % ERROR_THRESHOLD == 1;
}
}
Original file line number Diff line number Diff line change
Expand Up @@ -14,15 +14,26 @@
*/
package org.hyperledger.besu.ethereum.eth.sync.fastsync.worldstate;

import static org.hyperledger.besu.ethereum.eth.sync.StorageExceptionManager.canRetryOnError;
import static org.hyperledger.besu.ethereum.eth.sync.StorageExceptionManager.errorCountAtThreshold;
import static org.hyperledger.besu.ethereum.eth.sync.StorageExceptionManager.getRetryableErrorCounter;

import org.hyperledger.besu.ethereum.core.BlockHeader;
import org.hyperledger.besu.ethereum.eth.sync.worldstate.WorldDownloadState;
import org.hyperledger.besu.ethereum.worldstate.WorldStateStorage;
import org.hyperledger.besu.ethereum.worldstate.WorldStateStorage.Updater;
import org.hyperledger.besu.plugin.services.exception.StorageException;
import org.hyperledger.besu.services.tasks.Task;

import java.util.List;

import org.slf4j.Logger;
import org.slf4j.LoggerFactory;

public class PersistDataStep {

private static final Logger LOG = LoggerFactory.getLogger(PersistDataStep.class);

private final WorldStateStorage worldStateStorage;

public PersistDataStep(final WorldStateStorage worldStateStorage) {
Expand All @@ -33,24 +44,40 @@ public List<Task<NodeDataRequest>> persist(
final List<Task<NodeDataRequest>> tasks,
final BlockHeader blockHeader,
final WorldDownloadState<NodeDataRequest> downloadState) {
final Updater updater = worldStateStorage.updater();
tasks.stream()
.map(
task -> {
enqueueChildren(task, downloadState);
return task;
})
.map(Task::getData)
.filter(request -> request.getData() != null)
.forEach(
request -> {
if (isRootState(blockHeader, request)) {
downloadState.setRootNodeData(request.getData());
} else {
request.persist(updater);
}
});
updater.commit();
try {
final Updater updater = worldStateStorage.updater();
tasks.stream()
.map(
task -> {
enqueueChildren(task, downloadState);
return task;
})
.map(Task::getData)
.filter(request -> request.getData() != null)
.forEach(
request -> {
if (isRootState(blockHeader, request)) {
downloadState.setRootNodeData(request.getData());
} else {
request.persist(updater);
}
});
updater.commit();
} catch (StorageException storageException) {
if (canRetryOnError(storageException)) {
// We reset the task by setting it to null. This way, it is considered as failed by the
// pipeline, and it will attempt to execute it again later.
if (errorCountAtThreshold()) {
LOG.info(
"Encountered {} retryable RocksDB errors, latest error message {}",
getRetryableErrorCounter(),
storageException.getMessage());
}
tasks.forEach(nodeDataRequestTask -> nodeDataRequestTask.getData().setData(null));
} else {
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

can we not fail the future for an individual task in the foreach rather than failing the whole list? or is that a lot more plumbing to do in the pipeline to handle a single failed task future?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

not possible because all modification are in the same updater and we cannot commit only a part of a transaction

throw storageException;
}
}
return tasks;
}

Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -14,11 +14,16 @@
*/
package org.hyperledger.besu.ethereum.eth.sync.snapsync;

import static org.hyperledger.besu.ethereum.eth.sync.StorageExceptionManager.canRetryOnError;
import static org.hyperledger.besu.ethereum.eth.sync.StorageExceptionManager.errorCountAtThreshold;
import static org.hyperledger.besu.ethereum.eth.sync.StorageExceptionManager.getRetryableErrorCounter;

import org.hyperledger.besu.ethereum.eth.sync.snapsync.request.SnapDataRequest;
import org.hyperledger.besu.ethereum.eth.sync.snapsync.request.heal.TrieNodeHealingRequest;
import org.hyperledger.besu.ethereum.worldstate.WorldStateStorage;
import org.hyperledger.besu.metrics.BesuMetricCategory;
import org.hyperledger.besu.plugin.services.MetricsSystem;
import org.hyperledger.besu.plugin.services.exception.StorageException;
import org.hyperledger.besu.plugin.services.metrics.Counter;
import org.hyperledger.besu.services.pipeline.Pipe;
import org.hyperledger.besu.services.tasks.Task;
Expand All @@ -27,9 +32,12 @@
import java.util.stream.Stream;

import org.apache.tuweni.bytes.Bytes;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;

public class LoadLocalDataStep {

private static final Logger LOG = LoggerFactory.getLogger(LoadLocalDataStep.class);
private final WorldStateStorage worldStateStorage;
private final SnapWorldDownloadState downloadState;
private final SnapSyncProcessState snapSyncState;
Expand Down Expand Up @@ -58,19 +66,35 @@ public Stream<Task<SnapDataRequest>> loadLocalDataTrieNode(
final Task<SnapDataRequest> task, final Pipe<Task<SnapDataRequest>> completedTasks) {
final TrieNodeHealingRequest request = (TrieNodeHealingRequest) task.getData();
// check if node is already stored in the worldstate
if (snapSyncState.hasPivotBlockHeader()) {
Optional<Bytes> existingData = request.getExistingData(downloadState, worldStateStorage);
if (existingData.isPresent()) {
existingNodeCounter.inc();
request.setData(existingData.get());
request.setRequiresPersisting(false);
final WorldStateStorage.Updater updater = worldStateStorage.updater();
request.persist(
worldStateStorage, updater, downloadState, snapSyncState, snapSyncConfiguration);
updater.commit();
downloadState.enqueueRequests(request.getRootStorageRequests(worldStateStorage));
completedTasks.put(task);
return Stream.empty();
try {
if (snapSyncState.hasPivotBlockHeader()) {
Optional<Bytes> existingData = request.getExistingData(downloadState, worldStateStorage);
if (existingData.isPresent()) {
existingNodeCounter.inc();
request.setData(existingData.get());
request.setRequiresPersisting(false);
final WorldStateStorage.Updater updater = worldStateStorage.updater();
request.persist(
worldStateStorage, updater, downloadState, snapSyncState, snapSyncConfiguration);
updater.commit();
downloadState.enqueueRequests(request.getRootStorageRequests(worldStateStorage));
completedTasks.put(task);
return Stream.empty();
}
}
} catch (StorageException storageException) {
if (canRetryOnError(storageException)) {
// We reset the task by setting it to null. This way, it is considered as failed by the
// pipeline, and it will attempt to execute it again later.
if (errorCountAtThreshold()) {
LOG.info(
"Encountered {} retryable RocksDB errors, latest error message {}",
getRetryableErrorCounter(),
storageException.getMessage());
}
task.getData().clear();
} else {
throw storageException;
}
}
return Stream.of(task);
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -14,16 +14,25 @@
*/
package org.hyperledger.besu.ethereum.eth.sync.snapsync;

import static org.hyperledger.besu.ethereum.eth.sync.StorageExceptionManager.canRetryOnError;
import static org.hyperledger.besu.ethereum.eth.sync.StorageExceptionManager.errorCountAtThreshold;
import static org.hyperledger.besu.ethereum.eth.sync.StorageExceptionManager.getRetryableErrorCounter;

import org.hyperledger.besu.ethereum.bonsai.storage.BonsaiWorldStateKeyValueStorage;
import org.hyperledger.besu.ethereum.eth.sync.snapsync.request.SnapDataRequest;
import org.hyperledger.besu.ethereum.eth.sync.snapsync.request.heal.TrieNodeHealingRequest;
import org.hyperledger.besu.ethereum.worldstate.WorldStateStorage;
import org.hyperledger.besu.plugin.services.exception.StorageException;
import org.hyperledger.besu.services.tasks.Task;

import java.util.List;
import java.util.stream.Stream;

import org.slf4j.Logger;
import org.slf4j.LoggerFactory;

public class PersistDataStep {
private static final Logger LOG = LoggerFactory.getLogger(PersistDataStep.class);

private final SnapSyncProcessState snapSyncState;
private final WorldStateStorage worldStateStorage;
Expand All @@ -43,41 +52,58 @@ public PersistDataStep(
}

public List<Task<SnapDataRequest>> persist(final List<Task<SnapDataRequest>> tasks) {
final WorldStateStorage.Updater updater = worldStateStorage.updater();
for (Task<SnapDataRequest> task : tasks) {
if (task.getData().isResponseReceived()) {
// enqueue child requests
final Stream<SnapDataRequest> childRequests =
task.getData().getChildRequests(downloadState, worldStateStorage, snapSyncState);
if (!(task.getData() instanceof TrieNodeHealingRequest)) {
enqueueChildren(childRequests);
} else {
if (!task.getData().isExpired(snapSyncState)) {
try {
final WorldStateStorage.Updater updater = worldStateStorage.updater();
for (Task<SnapDataRequest> task : tasks) {
if (task.getData().isResponseReceived()) {
// enqueue child requests
final Stream<SnapDataRequest> childRequests =
task.getData().getChildRequests(downloadState, worldStateStorage, snapSyncState);
if (!(task.getData() instanceof TrieNodeHealingRequest)) {
enqueueChildren(childRequests);
} else {
continue;
if (!task.getData().isExpired(snapSyncState)) {
enqueueChildren(childRequests);
} else {
continue;
}
}
}

// persist nodes
final int persistedNodes =
task.getData()
.persist(
worldStateStorage,
updater,
downloadState,
snapSyncState,
snapSyncConfiguration);
if (persistedNodes > 0) {
if (task.getData() instanceof TrieNodeHealingRequest) {
downloadState.getMetricsManager().notifyTrieNodesHealed(persistedNodes);
} else {
downloadState.getMetricsManager().notifyNodesGenerated(persistedNodes);
// persist nodes
final int persistedNodes =
task.getData()
.persist(
worldStateStorage,
updater,
downloadState,
snapSyncState,
snapSyncConfiguration);
if (persistedNodes > 0) {
if (task.getData() instanceof TrieNodeHealingRequest) {
downloadState.getMetricsManager().notifyTrieNodesHealed(persistedNodes);
} else {
downloadState.getMetricsManager().notifyNodesGenerated(persistedNodes);
}
}
}
}
updater.commit();
} catch (StorageException storageException) {
if (canRetryOnError(storageException)) {
// We reset the task by setting it to null. This way, it is considered as failed by the
// pipeline, and it will attempt to execute it again later. not display all the retryable
// issues
if (errorCountAtThreshold()) {
LOG.info(
"Encountered {} retryable RocksDB errors, latest error message {}",
getRetryableErrorCounter(),
storageException.getMessage());
}
tasks.forEach(task -> task.getData().clear());
} else {
throw storageException;
}
}
updater.commit();
return tasks;
}

Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -231,6 +231,22 @@ public SnapWorldStateDownloadProcess build() {
"step",
"action");

/*
The logic and intercommunication of different pipelines can be summarized as follows:
1. Account Data Pipeline (fetchAccountDataPipeline): This process starts with downloading the leaves of the account tree in ranges, with multiple ranges being processed simultaneously.
If the downloaded accounts are smart contracts, tasks are created in the storage pipeline to download the storage tree of the smart contract, and in the code download pipeline for the smart contract.
2. Storage Data Pipeline (fetchStorageDataPipeline): Running parallel to the account data pipeline, this pipeline downloads the storage of smart contracts.
If all slots cannot be downloaded at once, tasks are created in the fetchLargeStorageDataPipeline to download the storage by range, allowing parallelization of large account downloads.
3. Code Data Pipeline (fetchCodePipeline): This pipeline, running concurrently with the account and storage data pipelines, is responsible for downloading the code of the smart contracts.
4. Large Storage Data Pipeline (fetchLargeStorageDataPipeline): This pipeline is used when the storage data for a smart contract is too large to be downloaded at once.
It enables the storage data to be downloaded in ranges, similar to the account data.
5. Healing Phase: Initiated after all other pipelines have completed their tasks, this phase ensures the integrity and completeness of the downloaded data.
*/
Comment on lines +234 to +249
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

🙏

final Pipeline<Task<SnapDataRequest>> completionPipeline =
PipelineBuilder.<Task<SnapDataRequest>>createPipeline(
"requestDataAvailable", bufferCapacity, outputCounter, true, "node_data_request")
Expand Down
Loading
Loading