Signal/Signal-Server
1
0
Fork 0
mirror of https://github.com/signalapp/Signal-Server synced 2026-04-20 07:08:05 +01:00
Code Issues Packages Projects Releases Wiki Activity

Add asynchronous chunk pre-read to AccountDatabaseCrawler

Browse Source
This commit is contained in:
Chris Eager
2021-09-08 10:32:30 -07:00
committed by Chris Eager
parent 624e40e3b7
commit b91a69d8b3
6 changed files with 89 additions and 51 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
service/src/main/java/org/whispersystems/textsecuregcm/WhisperServerService.java
View File

@@ -400,7 +400,10 @@ public class WhisperServerService extends Application<WhisperServerConfiguration
ExecutorService backupServiceExecutor = environment.lifecycle().executorService(name(getClass(), "backupService-%d")).maxThreads(1).minThreads(1).build();
ExecutorService storageServiceExecutor = environment.lifecycle().executorService(name(getClass(), "storageService-%d")).maxThreads(1).minThreads(1).build();
ExecutorService donationExecutor = environment.lifecycle().executorService(name(getClass(), "donation-%d")).maxThreads(1).minThreads(1).build();
ExecutorService multiRecipientMessageExecutor = environment.lifecycle().executorService(name(getClass(), "multiRecipientMessage-%d")).minThreads(64).maxThreads(64).build();
ExecutorService multiRecipientMessageExecutor = environment.lifecycle()
.executorService(name(getClass(), "multiRecipientMessage-%d")).minThreads(64).maxThreads(64).build();
ExecutorService accountsCrawlerChunkPreReadExecutor = environment.lifecycle()
.executorService(name(getClass(), "accountsCrawler-%d")).maxThreads(2).minThreads(2).build();
ExternalServiceCredentialGenerator directoryCredentialsGenerator = new ExternalServiceCredentialGenerator(config.getDirectoryConfiguration().getDirectoryClientConfiguration().getUserAuthenticationTokenSharedSecret(),
config.getDirectoryConfiguration().getDirectoryClientConfiguration().getUserAuthenticationTokenUserIdSecret(),
@@ -504,6 +507,7 @@ public class WhisperServerService extends Application<WhisperServerConfiguration
accountDatabaseCrawlerCache, accountDatabaseCrawlerListeners,
config.getAccountDatabaseCrawlerConfiguration().getChunkSize(),
config.getAccountDatabaseCrawlerConfiguration().getChunkIntervalMs(),
accountsCrawlerChunkPreReadExecutor,
dynamicConfigurationManager);
AccountDatabaseCrawlerCache dynamoDbMigrationCrawlerCache = new AccountDatabaseCrawlerCache(cacheCluster);
@@ -513,6 +517,7 @@ public class WhisperServerService extends Application<WhisperServerConfiguration
List.of(new AccountsDynamoDbMigrator(accountsDynamoDb, dynamicConfigurationManager)),
config.getDynamoDbMigrationCrawlerConfiguration().getChunkSize(),
config.getDynamoDbMigrationCrawlerConfiguration().getChunkIntervalMs(),
accountsCrawlerChunkPreReadExecutor,
dynamicConfigurationManager);
DeletedAccountsTableCrawler deletedAccountsTableCrawler = new DeletedAccountsTableCrawler(deletedAccountsManager, deletedAccountsDirectoryReconcilers, cacheCluster, recurringJobExecutor);

7
service/src/main/java/org/whispersystems/textsecuregcm/configuration/dynamic/DynamicAccountsDynamoDbMigrationConfiguration.java
View File

@@ -23,6 +23,9 @@ public class DynamicAccountsDynamoDbMigrationConfiguration {
@JsonProperty
boolean logMismatches;
@JsonProperty
boolean crawlerPreReadNextChunkEnabled;
@JsonProperty
boolean dynamoCrawlerEnabled;
@@ -71,6 +74,10 @@ public class DynamicAccountsDynamoDbMigrationConfiguration {
return logMismatches;
}
public boolean isCrawlerPreReadNextChunkEnabled() {
return crawlerPreReadNextChunkEnabled;
}
public boolean isDynamoCrawlerEnabled() {
return dynamoCrawlerEnabled;
}

51
service/src/main/java/org/whispersystems/textsecuregcm/storage/AccountDatabaseCrawler.java
View File

@@ -14,6 +14,7 @@ import io.dropwizard.lifecycle.Managed;
import java.util.List;
import java.util.Optional;
import java.util.UUID;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.atomic.AtomicBoolean;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
@@ -26,6 +27,8 @@ public class AccountDatabaseCrawler implements Managed, Runnable {
private static final Logger logger = LoggerFactory.getLogger(AccountDatabaseCrawler.class);
private static final MetricRegistry metricRegistry = SharedMetricRegistries.getOrCreate(Constants.METRICS_NAME);
private static final Timer readChunkTimer = metricRegistry.timer(name(AccountDatabaseCrawler.class, "readChunk"));
private static final Timer preReadChunkTimer = metricRegistry.timer(
name(AccountDatabaseCrawler.class, "preReadChunk"));
private static final Timer processChunkTimer = metricRegistry.timer(
name(AccountDatabaseCrawler.class, "processChunk"));
@@ -38,6 +41,7 @@ public class AccountDatabaseCrawler implements Managed, Runnable {
private final String workerId;
private final AccountDatabaseCrawlerCache cache;
private final List<AccountDatabaseCrawlerListener> listeners;
private final ExecutorService chunkPreReadExecutorService;
private final DynamicConfigurationManager dynamicConfigurationManager;
@@ -45,18 +49,19 @@ public class AccountDatabaseCrawler implements Managed, Runnable {
private boolean finished;
public AccountDatabaseCrawler(AccountsManager accounts,
AccountDatabaseCrawlerCache cache,
List<AccountDatabaseCrawlerListener> listeners,
int chunkSize,
long chunkIntervalMs,
DynamicConfigurationManager dynamicConfigurationManager)
{
this.accounts = accounts;
this.chunkSize = chunkSize;
this.chunkIntervalMs = chunkIntervalMs;
this.workerId = UUID.randomUUID().toString();
this.cache = cache;
this.listeners = listeners;
AccountDatabaseCrawlerCache cache,
List<AccountDatabaseCrawlerListener> listeners,
int chunkSize,
long chunkIntervalMs,
ExecutorService chunkPreReadExecutorService,
DynamicConfigurationManager dynamicConfigurationManager) {
this.accounts = accounts;
this.chunkSize = chunkSize;
this.chunkIntervalMs = chunkIntervalMs;
this.workerId = UUID.randomUUID().toString();
this.cache = cache;
this.listeners = listeners;
this.chunkPreReadExecutorService = chunkPreReadExecutorService;
this.dynamicConfigurationManager = dynamicConfigurationManager;
}
@@ -136,6 +141,8 @@ public class AccountDatabaseCrawler implements Managed, Runnable {
final AccountCrawlChunk chunkAccounts = readChunk(fromUuid, chunkSize, useDynamo);
primeDatabaseForNextChunkAsync(chunkAccounts.getLastUuid(), chunkSize, useDynamo);
if (chunkAccounts.getAccounts().isEmpty()) {
logger.info("Finished crawl");
listeners.forEach(listener -> listener.onCrawlEnd(fromUuid));
@@ -156,8 +163,26 @@ public class AccountDatabaseCrawler implements Managed, Runnable {
}
}
/**
* This is an optimization based on the observation that cold reads of chunks are slow, but subsequent reads of the
* same chunk (within a few minutes) are fast. We can’t easily store the actual result data, since the next chunk
* might be processed elsewhere, but the time savings are still substantial.
*/
private void primeDatabaseForNextChunkAsync(Optional<UUID> fromUuid, int chunkSize, boolean useDynamo) {
if (dynamicConfigurationManager.getConfiguration().getAccountsDynamoDbMigrationConfiguration()
.isCrawlerPreReadNextChunkEnabled()) {
if (!useDynamo && fromUuid.isPresent()) {
chunkPreReadExecutorService.submit(() -> readChunk(fromUuid, chunkSize, false, preReadChunkTimer));
}
}
}
private AccountCrawlChunk readChunk(Optional<UUID> fromUuid, int chunkSize, boolean useDynamo) {
try (Timer.Context timer = readChunkTimer.time()) {
return readChunk(fromUuid, chunkSize, useDynamo, readChunkTimer);
}
private AccountCrawlChunk readChunk(Optional<UUID> fromUuid, int chunkSize, boolean useDynamo, Timer readTimer) {
try (Timer.Context timer = readTimer.time()) {
if (fromUuid.isPresent()) {
return useDynamo