[compat] [server] [client] [test] Global RT DIV: Chunking Support #1385
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -12,6 +12,7 @@ | |
| import static com.linkedin.venice.ConfigKeys.KAFKA_BOOTSTRAP_SERVERS; | ||
| import static com.linkedin.venice.LogMessages.KILLED_JOB_MESSAGE; | ||
| import static com.linkedin.venice.kafka.protocol.enums.ControlMessageType.START_OF_SEGMENT; | ||
| import static com.linkedin.venice.serialization.avro.AvroProtocolDefinition.GLOBAL_RT_DIV_STATE; | ||
| import static com.linkedin.venice.utils.Utils.FATAL_DATA_VALIDATION_ERROR; | ||
| import static com.linkedin.venice.utils.Utils.getReplicaId; | ||
| import static java.util.Comparator.comparingInt; | ||
|
|
@@ -48,6 +49,7 @@ | |
| import com.linkedin.venice.common.VeniceSystemStoreType; | ||
| import com.linkedin.venice.common.VeniceSystemStoreUtils; | ||
| import com.linkedin.venice.compression.CompressionStrategy; | ||
| import com.linkedin.venice.compression.NoopCompressor; | ||
| import com.linkedin.venice.compression.VeniceCompressor; | ||
| import com.linkedin.venice.exceptions.DiskLimitExhaustedException; | ||
| import com.linkedin.venice.exceptions.MemoryLimitExhaustedException; | ||
|
|
@@ -184,6 +186,8 @@ public abstract class StoreIngestionTask implements Runnable, Closeable { | |
| private static final int MAX_KILL_CHECKING_ATTEMPTS = 10; | ||
| private static final int CHUNK_MANIFEST_SCHEMA_ID = | ||
| AvroProtocolDefinition.CHUNKED_VALUE_MANIFEST.getCurrentProtocolVersion(); | ||
| private static final int GLOBAL_RT_DIV_STATE_SCHEMA_ID = | ||
| AvroProtocolDefinition.GLOBAL_RT_DIV_STATE.getCurrentProtocolVersion(); | ||
|
|
||
| protected static final RedundantExceptionFilter REDUNDANT_LOGGING_FILTER = | ||
| RedundantExceptionFilter.getRedundantExceptionFilter(); | ||
|
|
@@ -317,6 +321,7 @@ public abstract class StoreIngestionTask implements Runnable, Closeable { | |
| protected final IngestionNotificationDispatcher ingestionNotificationDispatcher; | ||
|
|
||
| protected final ChunkAssembler chunkAssembler; | ||
| protected final ChunkAssembler divChunkAssembler; | ||
| private final Optional<ObjectCacheBackend> cacheBackend; | ||
| private DaVinciRecordTransformer recordTransformer; | ||
|
|
||
|
|
@@ -467,6 +472,8 @@ public StoreIngestionTask( | |
| new IngestionNotificationDispatcher(notifiers, kafkaVersionTopic, isCurrentVersion); | ||
| this.missingSOPCheckExecutor.execute(() -> waitForStateVersion(kafkaVersionTopic)); | ||
| this.chunkAssembler = new ChunkAssembler(storeName); | ||
| this.divChunkAssembler = | ||
| builder.getDivChunkAssembler() != null ? builder.getDivChunkAssembler() : new ChunkAssembler(storeName); | ||
|
There was a problem hiding this comment. It seems |
||
| this.cacheBackend = cacheBackend; | ||
|
|
||
| if (recordTransformerFunction != null) { | ||
|
|
@@ -1145,6 +1152,12 @@ private int handleSingleMessage( | |
| record.getTopicPartition().getPartitionNumber(), | ||
| partitionConsumptionStateMap.get(topicPartition.getPartitionNumber())); | ||
| } | ||
| } else if (record.getKey().isGlobalRtDiv()) { | ||
| // TODO: This is a placeholder for the actual implementation. | ||
| if (isGlobalRtDivEnabled) { | ||
| processGlobalRtDivMessage(record); // This is a global realtime topic data integrity validator snapshot | ||
| } | ||
| return 0; | ||
| } | ||
|
|
||
| // This function may modify the original record in KME and it is unsafe to use the payload from KME directly after | ||
|
|
@@ -1189,6 +1202,28 @@ private int handleSingleMessage( | |
| return record.getPayloadSize(); | ||
| } | ||
|
|
||
| void processGlobalRtDivMessage(PubSubMessage<KafkaKey, KafkaMessageEnvelope, Long> record) { | ||
| KafkaKey key = record.getKey(); | ||
| KafkaMessageEnvelope value = record.getValue(); | ||
| Put put = (Put) value.getPayloadUnion(); | ||
|
|
||
| Object assembledObject = divChunkAssembler.bufferAndAssembleRecord( | ||
|
There was a problem hiding this comment. I saw this function is using in-memory storage engine as a temp storage, which will increase the memory usage, which is different from the regular data chunk handling, which would minimize the memory usage, and I think this is also what we agreed during the design review. |
||
| record.getTopicPartition(), | ||
| put.getSchemaId(), | ||
| key.getKey(), | ||
| put.getPutValue(), | ||
| record.getOffset(), | ||
| put.getSchemaId(), | ||
| new NoopCompressor(), | ||
|
There was a problem hiding this comment. Shall we always use a certain compression algo for |
||
| (valueBytes) -> GLOBAL_RT_DIV_STATE.getSerializer() | ||
| .deserialize(ByteUtils.extractByteArray(valueBytes), GLOBAL_RT_DIV_STATE_SCHEMA_ID)); | ||
|
There was a problem hiding this comment. Shall we update |
||
|
|
||
| if (assembledObject == null) { | ||
| return; // the message value only contained one data chunk, so the Global RT DIV cannot yet be fully assembled | ||
| } | ||
| // TODO: We will add the code to process Global RT DIV message later in here. | ||
| } | ||
|
|
||
| /** | ||
| * This function is in charge of producing the consumer records to the writer buffers maintained by {@link StoreBufferService}. | ||
| * | ||
|
|
@@ -2397,6 +2432,12 @@ protected boolean shouldProcessRecord(PubSubMessage<KafkaKey, KafkaMessageEnvelo | |
| return false; | ||
| } | ||
|
|
||
| // Global RT DIV messages should only be in version topics, not realtime topics. Skip it and log a warning. | ||
| if (record.getKey().isGlobalRtDiv() && record.getTopic().isRealTime()) { | ||
|
There was a problem hiding this comment. What are the scenarios that the RT topic would contain |
||
| LOGGER.warn("Skipping Global RT DIV message from realtime topic partition: {}", record.getTopicPartition()); | ||
| return false; | ||
| } | ||
|
|
||
| if (partitionConsumptionState.isEndOfPushReceived() && partitionConsumptionState.isBatchOnly()) { | ||
| KafkaKey key = record.getKey(); | ||
| KafkaMessageEnvelope value = record.getValue(); | ||
|
|
@@ -3786,7 +3827,7 @@ private int processKafkaDataMessage( | |
| private void waitReadyToProcessRecord(PubSubMessage<KafkaKey, KafkaMessageEnvelope, Long> record) | ||
| throws InterruptedException { | ||
| KafkaMessageEnvelope kafkaValue = record.getValue(); | ||
| if (record.getKey().isControlMessage() || record.getKey().isGlobalRtDiv() || kafkaValue == null) { | ||
|
There was a problem hiding this comment. Can |
||
| return; | ||
| } | ||
|
|
||
|
|
@@ -4561,4 +4602,8 @@ void setVersionRole(PartitionReplicaIngestionContext.VersionRole versionRole) { | |
| protected boolean isDaVinciClient() { | ||
| return isDaVinciClient; | ||
| } | ||
|
|
||
| ChunkAssembler getDivChunkAssembler() { | ||
| return this.divChunkAssembler; | ||
| } | ||
| } | ||
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Oops, something went wrong.
Oops, something went wrong.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Why do you do this change? Personally I think this makes less explicit about the offset map we are tracking
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Because I feel that
reset()is not the right word for it. It's not clearing the state or replacing the map entirely. It's updating the existing map with the values of the new map and retaining the values missing from the new map.I can rename it to
updateUpstreamOffsets().