Signal/Signal-Server
1
0
Fork 0
mirror of https://github.com/signalapp/Signal-Server synced 2026-04-20 11:48:07 +01:00
Code Issues Packages Projects Releases Wiki Activity

Use queries instead of scans when fetching experiment samples (again)

Browse Source
This commit is contained in:
Jon Chambers
2024-08-07 17:09:42 -04:00
committed by Jon Chambers
parent f682af2fe0
commit 206e97d374
4 changed files with 15 additions and 62 deletions
Download Patch File Download Diff File Expand all files Collapse all files

36
service/src/main/java/org/whispersystems/textsecuregcm/experiment/PushNotificationExperimentSamples.java
View File

@@ -14,7 +14,6 @@ import org.whispersystems.textsecuregcm.util.ExceptionUtils;
import org.whispersystems.textsecuregcm.util.SystemMapper;
import reactor.core.publisher.Flux;
import reactor.core.publisher.Mono;
import reactor.core.scheduler.Scheduler;
import reactor.util.function.Tuple2;
import reactor.util.function.Tuples;
import reactor.util.retry.Retry;
@@ -24,6 +23,7 @@ import software.amazon.awssdk.services.dynamodb.model.AttributeValue;
import software.amazon.awssdk.services.dynamodb.model.ConditionalCheckFailedException;
import software.amazon.awssdk.services.dynamodb.model.DeleteItemRequest;
import software.amazon.awssdk.services.dynamodb.model.PutItemRequest;
import software.amazon.awssdk.services.dynamodb.model.QueryRequest;
import software.amazon.awssdk.services.dynamodb.model.ReturnValue;
import software.amazon.awssdk.services.dynamodb.model.ScanRequest;
import software.amazon.awssdk.services.dynamodb.model.UpdateItemRequest;
@@ -198,7 +198,6 @@ public class PushNotificationExperimentSamples {
*
* @param experimentName the name of the experiment for which to fetch samples
* @param stateClass the type of state object for sample in the given experiment
* @param totalSegments the number of segments into which the scan of the backing data store will be divided
*
* @return a publisher of tuples of ACI, device ID, and sample for all samples associated with the given experiment
*
@@ -206,38 +205,11 @@ public class PushNotificationExperimentSamples {
*
* @see <a href="https://docs.aws.amazon.com/amazondynamodb/latest/developerguide/Scan.html#Scan.ParallelScan">Working with scans - Parallel scan</a>
*/
public <T> Flux<PushNotificationExperimentSample<T>> getSamples(final String experimentName,
final Class<T> stateClass,
final int totalSegments,
final Scheduler scheduler) {
public <T> Flux<PushNotificationExperimentSample<T>> getSamples(final String experimentName, final Class<T> stateClass) {
// Note that we're using a DynamoDB Scan operation instead of a Query. A Query would allow us to limit the search
// space to a specific experiment, but doesn't allow us to use segments. A Scan will always inspect all items in the
// table, but allows us to segment the search. Since we're generally calling this method in conjunction with "…and
// record a final state for the sample," distributing reads/writes across shards helps us avoid per-partition
// throughput limits. If we wind up with many concurrent experiments, it may be worthwhile to revisit this decision.
if (totalSegments < 1) {
throw new IllegalArgumentException("Total number of segments must be positive");
}
return Flux.range(0, totalSegments)
.parallel()
.runOn(scheduler)
.flatMap(segment -> getSamplesFromSegment(experimentName, stateClass, segment, totalSegments))
.sequential();
}
private <T> Flux<PushNotificationExperimentSample<T>> getSamplesFromSegment(final String experimentName,
final Class<T> stateClass,
final int segment,
final int totalSegments) {
return Flux.from(dynamoDbAsyncClient.scanPaginator(ScanRequest.builder()
return Flux.from(dynamoDbAsyncClient.queryPaginator(QueryRequest.builder()
.tableName(tableName)
.segment(segment)
.totalSegments(totalSegments)
.filterExpression("#experiment = :experiment")
.keyConditionExpression("#experiment = :experiment")
.expressionAttributeNames(Map.of("#experiment", KEY_EXPERIMENT_NAME))
.expressionAttributeValues(Map.of(":experiment", AttributeValue.fromS(experimentName)))
.build())