Package com.linkedin.venice.controller

Class VeniceHelixAdmin

java.lang.Object

com.linkedin.venice.controller.VeniceHelixAdmin

All Implemented Interfaces:

Admin, StoreCleaner, Closeable, AutoCloseable

public class VeniceHelixAdmin
extends Object
implements Admin, StoreCleaner

Helix Admin based on 0.8.4.215 APIs.

After using controller as service mode. There are two levels of cluster and controllers. Each venice controller will hold a level1 helix controller which will keep connecting to Helix, there is a cluster only used for all of these level1 controllers(controller's cluster). The second level is our venice clusters. Like prod cluster, dev cluster etc. Each of cluster will be Helix resource in the controller's cluster. Helix will choose one of level1 controller becoming the leader of our venice cluster. In our distributed controllers state transition handler, a level2 controller will be initialized to manage this venice cluster only. If this level1 controller is chosen as the leader controller of multiple Venice clusters, multiple level2 controller will be created based on cluster specific config.

Admin is shared by multiple cluster's controllers running in one physical Venice controller instance.

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static interface	VeniceHelixAdmin.StoreMetadataOperation

Nested classes/interfaces inherited from interface com.linkedin.venice.controller.Admin

Admin.OfflinePushStatusInfo

Field Summary

Fields

Modifier and Type	Field	Description
protected static final int	INTERNAL_STORE_GET_RRT_TOPIC_ATTEMPTS
protected static final long	INTERNAL_STORE_RTT_RETRY_BACKOFF_MS
protected final PubSubTopicRepository	pubSubTopicRepository

Constructor Summary

Constructors

Constructor	Description
VeniceHelixAdmin(VeniceControllerMultiClusterConfig multiClusterConfigs, io.tehuti.metrics.MetricsRepository metricsRepository, boolean sslEnabled, com.linkedin.d2.balancer.D2Client d2Client, Optional<SSLConfig> sslConfig, Optional<DynamicAccessController> accessController, Optional<ICProvider> icProvider, PubSubTopicRepository pubSubTopicRepository, PubSubClientsFactory pubSubClientsFactory, List<ClusterLeaderInitializationRoutine> additionalInitRoutines)
VeniceHelixAdmin(VeniceControllerMultiClusterConfig multiClusterConfigs, io.tehuti.metrics.MetricsRepository metricsRepository, com.linkedin.d2.balancer.D2Client d2Client, PubSubTopicRepository pubSubTopicRepository, PubSubClientsFactory pubSubClientsFactory)

Method Summary

Modifier and Type	Method	Description
void	abortMigration(String srcClusterName, String destClusterName, String storeName)	Abort store migration by resetting migration flag at the source cluster, resetting storeConfig, and updating "cluster" in "/storeConfigs" znode back to the source cluster.
DerivedSchemaEntry	addDerivedSchema(String clusterName, String storeName, int valueSchemaId, int derivedSchemaId, String derivedSchemaStr)	Add a new derived schema for the given store with all specified properties.
DerivedSchemaEntry	addDerivedSchema(String clusterName, String storeName, int valueSchemaId, String derivedSchemaStr)	Add a new derived schema for the given store with all specified properties and return a new DerivedSchemaEntry object containing the schema and its id.
void	addInstanceToAllowlist(String clusterName, String helixNodeId)	Add the given helix nodeId into the allowlist in ZK.
RmdSchemaEntry	addReplicationMetadataSchema(String clusterName, String storeName, int valueSchemaId, int replicationMetadataVersionId, String replicationMetadataSchemaStr)	Create a new ReplicationMetadataSchemaEntry object with the given properties and add it into schema repository if no duplication.
boolean	addSpecificVersion(String clusterName, String storeName, Version version)	TODO refactor addVersion to these broken down methods instead of doing everything in one giant method.
SchemaEntry	addSupersetSchema(String clusterName, String storeName, String valueSchema, int valueSchemaId, String supersetSchemaStr, int supersetSchemaId)	Add a new superset schema for the given store with all specified properties.
SchemaEntry	addValueSchema(String clusterName, String storeName, String valueSchemaStr, int schemaId, DirectionalSchemaCompatibilityType compatibilityType)	Add a new value schema for the given store with all specified properties and return a new SchemaEntry object containing the schema and its id.
SchemaEntry	addValueSchema(String clusterName, String storeName, String valueSchemaStr, DirectionalSchemaCompatibilityType expectedCompatibilityType)
void	addVersionAndStartIngestion(String clusterName, String storeName, String pushJobId, int versionNumber, int numberOfPartitions, Version.PushType pushType, String remoteKafkaBootstrapServers, long rewindTimeInSecondsOverride, int replicationMetadataVersionId, boolean versionSwapDeferred, int repushSourceVersion)	This method behaves differently in VeniceHelixAdmin and VeniceParentHelixAdmin.
void	addVersionAndStartIngestion(String clusterName, String storeName, String pushJobId, int versionNumber, int numberOfPartitions, Version.PushType pushType, String remoteKafkaBootstrapServers, long rewindTimeInSecondsOverride, int replicationMetadataVersionId, boolean versionSwapDeferred, String targetedRegions, int repushSourceVersion)	This is a wrapper for VeniceHelixAdmin#addVersion but performs additional operations needed for add version invoked from the admin channel.
Pair<Boolean,Version>	addVersionAndTopicOnly(String clusterName, String storeName, String pushJobId, int versionNumber, int numberOfPartitions, int replicationFactor, boolean sendStartOfPush, boolean sorted, Version.PushType pushType, String compressionDictionary, String remoteKafkaBootstrapServers, Optional<String> sourceGridFabric, long rewindTimeInSecondsOverride, int replicationMetadataVersionId, Optional<String> emergencySourceRegion, boolean versionSwapDeferred)
Pair<Boolean,Version>	addVersionAndTopicOnly(String clusterName, String storeName, String pushJobId, int versionNumber, int numberOfPartitions, int replicationFactor, boolean sendStartOfPush, boolean sorted, Version.PushType pushType, String compressionDictionary, String remoteKafkaBootstrapServers, Optional<String> sourceGridFabric, long rewindTimeInSecondsOverride, int replicationMetadataVersionId, Optional<String> emergencySourceRegion, boolean versionSwapDeferred, String targetedRegions, int repushSourceVersion)	A wrapper to invoke VeniceHelixAdmin#addVersion to only increment the version and create the topic(s) needed without starting ingestion.
Version	addVersionOnly(String clusterName, String storeName, String pushJobId, int versionNumber, int numberOfPartitions, Version.PushType pushType, String remoteKafkaBootstrapServers, long rewindTimeInSecondsOverride, int replicationMetadataVersionId)	Only add version to the store without creating the topic or start ingestion.
int	calculateNumberOfPartitions(String clusterName, String storeName)	Calculate number of partition for given store.
protected void	checkPreConditionForCreateStore(String clusterName, String storeName, String keySchema, String valueSchema, boolean allowSystemStore, boolean skipLingeringResourceCheck)	Check whether Controller should block the incoming store creation.
void	checkResourceCleanupBeforeStoreCreation(String clusterName, String storeName)	Check whether there are any resource left for the store creation in cluster: If there is any, this function should throw Exception.
List<String>	cleanupInstanceCustomizedStates(String clusterName)	Scan through instance level customized states and remove any lingering ZNodes that are no longer relevant.
void	clearIngestionKillMessageAndVerify(String clusterName, String versionTopicName)	Clear KILL messages from a participant system store.
void	clearInstanceMonitor(String clusterName)
void	close()	Cause VeniceHelixAdmin and its associated services to stop executing.
StoreComparisonInfo	compareStore(String clusterName, String storeName, String fabricA, String fabricB)	Compare store metadata and version states between two fabrics.
void	completeMigration(String srcClusterName, String destClusterName, String storeName)
void	configureActiveActiveReplication(String clusterName, VeniceUserStoreType storeType, Optional<String> storeName, boolean enableActiveActiveReplicationForCluster, Optional<String> regionsFilter)	Enable/disable active active replications for certain stores (batch only, hybrid only, incremental push, hybrid or incremental push, all) in a cluster.
boolean	containsHelixResource(String clusterName, String kafkaTopic)	This purpose of the function is to check if the given resource exists in the Helix cluster.
StoreInfo	copyOverStoreSchemasAndConfigs(String clusterName, String srcFabric, String destFabric, String storeName)
void	createHelixResourceAndStartMonitoring(String clusterName, String storeName, Version version)	Create Helix-resources for a given storage node cluster and starts monitoring a new push.
void	createSpecificVersionTopic(String clusterName, String storeName, Version version)	Create the corresponding version topic based on the provided Version
void	createStoragePersona(String clusterName, String name, long quotaNumber, Set<String> storesToEnforce, Set<String> owners)
void	createStore(String clusterName, String storeName, String owner, String keySchema, String valueSchema, boolean isSystemStore, Optional<String> accessPermissions)	Create a new ZK store and its configuration in the store repository and create schemas in the schema repository.
void	deleteAclForStore(String clusterName, String storeName)	Delete the current set of ACL provisioned for a venice store and its associated kafka topic.
List<Version>	deleteAllVersionsInStore(String clusterName, String storeName)	Delete all of venice versions in given store(including venice resource, kafka topic, offline pushs and all related resources).
void	deleteHelixResource(String clusterName, String kafkaTopic)	This purpose of the function is to delete the given resource from the Helix cluster.
void	deleteOldVersionInStore(String clusterName, String storeName, int versionNum)	Delete the given version from the store.
void	deleteOneStoreVersion(String clusterName, String storeName, int versionNumber)	Delete version from cluster, removing all related resources
void	deleteParticipantStoreKillMessage(String clusterName, String kafkaTopic)	Compose a ParticipantMessageKey message and execute a delete operation on the key to the cluster's participant store.
void	deleteStoragePersona(String clusterName, String name)
void	deleteStore(String clusterName, String storeName, int largestUsedVersionNumber, boolean waitOnRTTopicDeletion)	This method will delete store data, metadata, version and rt topics One exception is for stores with isMigrating flag set.
void	deleteValueSchemas(String clusterName, String storeName, Set<Integer> unusedValueSchemaIds)	Deletes a store's values schema with ids `except` the ids passed in the argument inuseValueSchemaIds
Pair<String,String>	discoverCluster(String storeName)	Find the cluster which the given store belongs to.
void	enableDisabledPartition(String clusterName, String kafkaTopic, boolean enableAll)
Map<String,String>	findAllBootstrappingVersions(String clusterName)	Find the store versions which have at least one bootstrap replica.
String	getAclForStore(String clusterName, String storeName)	Fetch the current set of ACL provisioned for a venice store and its associated kafka topic.
HelixAdapterSerializer	getAdapterSerializer()
Optional<AdminCommandExecutionTracker>	getAdminCommandExecutionTracker(String clusterName)	Get the tracker used to track the execution of the admin command for the given cluster.
Map<String,Long>	getAdminTopicMetadata(String clusterName, Optional<String> storeName)
InstanceRemovableStatuses	getAggregatedHealthStatus(String cluster, List<String> instances, List<String> toBeStoppedInstances, boolean isSSLEnabled)
Optional<String>	getAggregateRealTimeTopicSource(String clusterName)	Return the source Kafka boostrap server url for aggregate real-time topic updates
Set<String>	getAllowlist(String clusterName)
Map<String,String>	getAllStorePushStrategyForMigration()	Unsupported operation in the child controller.
List<Store>	getAllStores(String clusterName)
Map<String,String>	getAllStoreStatuses(String clusterName)	Get the statuses of all stores.
int	getBackupVersion(String clusterName, String storeName)
long	getBackupVersionDefaultRetentionMs()	Returns default backup version retention time.
int	getBackupVersionNumber(List<Version> versions, int currentVersion)	Get backup version number, the largest online version number that is less than the current version number
Map<String,String>	getBackupVersionsForMultiColos(String clusterName, String storeName)
BatchJobHeartbeatValue	getBatchJobHeartbeatValue(BatchJobHeartbeatKey batchJobHeartbeatKey)
String	getChildControllerD2ServiceName(String clusterName)	Get child datacenter controller d2 service name
Map<String,String>	getChildDataCenterControllerD2Map(String clusterName)	Get child datacenter to child controller d2 zk host mapping
Map<String,String>	getChildDataCenterControllerUrlMap(String clusterName)	Get child datacenter to child controller url mapping.
List<String>	getClustersLeaderOf()	Get a list of clusters this controller is a leader of.
Map<String,StoreDataAudit>	getClusterStaleStores(String clusterName)
List<StoragePersona>	getClusterStoragePersonas(String clusterName)
ArrayList<StoreInfo>	getClusterStores(String clusterName)	Return all stores in a cluster.
Map<String,ControllerClient>	getControllerClientMap(String clusterName)
int	getCurrentVersion(String clusterName, String storeName)
Map<String,Integer>	getCurrentVersionsForMultiColos(String clusterName, String storeName)
int	getDefaultMaxRecordSizeBytes()
GeneratedSchemaID	getDerivedSchemaId(String clusterName, String storeName, String schemaStr)
Collection<DerivedSchemaEntry>	getDerivedSchemas(String clusterName, String storeName)
DisabledPartitionStats	getDisabledPartitionStats(String clusterName)
Optional<String>	getEmergencySourceRegion(String clusterName)	Return the emergency source region configuration.
ExecutionIdAccessor	getExecutionIdAccessor()
int	getFutureVersion(String clusterName, String storeName)
Map<String,String>	getFutureVersionsForMultiColos(String clusterName, String storeName)
long	getHeartbeatFromSystemStore(String clusterName, String systemStoreName)	Read the latest heartbeat timestamp from system store.
protected org.apache.helix.HelixAdmin	getHelixAdmin()
HelixAdminClient	getHelixAdminClient()
HelixVeniceClusterResources	getHelixVeniceClusterResources(String cluster)
Version	getIncrementalPushVersion(String clusterName, String storeName, String pushJobId)
Set<Integer>	getInUseValueSchemaIds(String clusterName, String storeName)
String	getKafkaBootstrapServers(boolean isSSL)	Return the ssl or non-ssl bootstrap servers based on the given flag.
SchemaEntry	getKeySchema(String clusterName, String storeName)
int	getLargestUsedVersionFromStoreGraveyard(String clusterName, String storeName)
Long	getLastSucceededExecutionId(String clusterName, String storeName)	Get last succeeded execution id for a given store; if storeName is null, return the last succeeded execution id for the cluster
Long	getLastSucceedExecutionId(String clusterName)	Get the id of the last succeed execution in this controller.
Instance	getLeaderController(String clusterName)	Get instance of leader controller.
LiveInstanceMonitor	getLiveInstanceMonitor(String clusterName)
MetaStoreReader	getMetaStoreReader()
StoreMetaValue	getMetaStoreValue(StoreMetaKey metaKey, String storeName)
MetaStoreWriter	getMetaStoreWriter()	Return a MetaStoreWriter, which can be shared across different Venice clusters.
int	getMinNumberOfUnusedKafkaTopicsToPreserve()
String	getNativeReplicationKafkaBootstrapServerAddress(String sourceFabric)
String	getNativeReplicationSourceFabric(String clusterName, Store store, Optional<String> sourceGridFabric, Optional<String> emergencySourceRegion, String targetedRegions)	Source fabric selection priority: 1.
Admin.OfflinePushStatusInfo	getOffLinePushStatus(String clusterName, String kafkaTopic)	Query the status of the offline push by given kafka topic.
Admin.OfflinePushStatusInfo	getOffLinePushStatus(String clusterName, String kafkaTopic, Optional<String> incrementalPushVersion, String region, String targetedRegions)
int	getOnlineFutureVersion(String clusterName, String storeName)
protected static ExecutionStatus	getOverallPushStatus(ExecutionStatus veniceStatus, ExecutionStatus daVinciStatus)
ParentControllerRegionState	getParentControllerRegionState()	Return the state of the region of the parent controller.
StoragePersona	getPersonaAssociatedWithStore(String clusterName, String storeName)
PubSubConsumerAdapterFactory	getPubSubConsumerAdapterFactory()
VeniceProperties	getPubSubSSLProperties(String pubSubBrokerAddress)
PubSubTopicRepository	getPubSubTopicRepository()
PushJobDetails	getPushJobDetails(PushJobStatusRecordKey key)
PushStatusStoreReader	getPushStatusStoreReader()
PushStatusStoreWriter	getPushStatusStoreWriter()
HelixReadOnlyZKSharedSchemaRepository	getReadOnlyZKSharedSchemaRepository()	Return a shared read only schema repository for zk shared stores.
HelixReadOnlyZKSharedSystemStoreRepository	getReadOnlyZKSharedSystemStoreRepository()	Return a shared read only store repository for zk shared stores.
Version	getReferenceVersionForStreamingWrites(String clusterName, String storeName, String pushJobId)
String	getRegionName()	Return the region name of this Admin
RegionPushDetails	getRegionPushDetails(String clusterName, String storeName, boolean isPartitionDetailAdded)
List<Replica>	getReplicas(String clusterName, String kafkaTopic)
List<Replica>	getReplicasOfStorageNode(String cluster, String instanceId)
int	getReplicationFactor(String clusterName, String storeName)
Optional<org.apache.avro.Schema>	getReplicationMetadataSchema(String clusterName, String storeName, int valueSchemaID, int rmdVersionID)
Collection<RmdSchemaEntry>	getReplicationMetadataSchemas(String clusterName, String storeName)
RepushInfo	getRepushInfo(String clusterName, String storeName, Optional<String> fabricName)
RoutersClusterConfig	getRoutersClusterConfig(String clusterName)	Get the cluster level config for all routers.
String	getServerD2Service(String clusterName)	Find the server d2 service associated with a given cluster name.
Optional<SSLFactory>	getSslFactory()
protected static Optional<Version>	getStartedVersion(Store store)	The intended semantic is to use this method to find the version that something is currently pushing to.
double	getStorageEngineOverheadRatio(String clusterName)
List<String>	getStorageNodes(String clusterName)
Map<String,String>	getStorageNodesStatus(String clusterName, boolean enableReplica)
StorageNodeStatus	getStorageNodesStatus(String clusterName, String instanceId)	Query and return the current status of the given storage node.
StoragePersona	getStoragePersona(String clusterName, String name)
Store	getStore(String clusterName, String storeName)
HelixReadOnlyStoreConfigRepository	getStoreConfigRepo()	Return a shared store config repository.
StoreGraveyard	getStoreGraveyard()
TopicManager	getTopicManager()
TopicManager	getTopicManager(String pubSubServerAddress)
SchemaEntry	getValueSchema(String clusterName, String storeName, int id)
int	getValueSchemaId(String clusterName, String storeName, String valueSchemaStr)
Collection<SchemaEntry>	getValueSchemas(String clusterName, String storeName)
VeniceWriterFactory	getVeniceWriterFactory()
org.apache.helix.zookeeper.impl.client.ZkClient	getZkClient()
boolean	hasStore(String clusterName, String storeName)	Test if the input store exists in a cluster.
Version	incrementVersionIdempotent(String clusterName, String storeName, String pushJobId, int numberOfPartitions, int replicationFactor, Version.PushType pushType, boolean sendStartOfPush, boolean sorted, String compressionDictionary, Optional<String> sourceGridFabric, Optional<X509Certificate> requesterCert, long rewindTimeInSecondsOverride, Optional<String> emergencySourceRegion, boolean versionSwapDeferred, String targetedRegions, int repushSourceVersion)	Note: this currently use the pushID to guarantee idempotence, unexpected behavior may result if multiple batch jobs push to the same store at the same time.
void	initiateDataRecovery(String clusterName, String storeName, int version, String sourceFabric, String destinationFabric, boolean copyAllVersionConfigs, Optional<Version> sourceFabricVersion)	Initiate data recovery for a store version given a source fabric.
void	initStorageCluster(String clusterName)	Create and configure the Venice storage cluster with required properties in Helix and waits the resource's (partial) partition to appear in the external view.
boolean	isActiveActiveReplicationEnabledInAllRegion(String clusterName, String storeName, boolean checkCurrentVersion)	Returns true if A/A replication is enabled in all child controller and parent controller.
boolean	isAdminTopicConsumptionEnabled(String clusterName)	Return whether the admin consumption task is enabled for the passed cluster.
boolean	isClusterValid(String clusterName)	Test if a cluster is valid (in Helix cluster list).
boolean	isClusterWipeAllowed(String clusterName)
NodeRemovableResult	isInstanceRemovable(String clusterName, String helixNodeId, List<String> lockedNodes)	Assuming all hosts identified by lockedNodes and their corresponding resources are unusable, is the given instance able to be removed out from the given cluster.
boolean	isLeaderControllerFor(String clusterName)	Check if this controller itself is the leader controller for a given cluster or not.
boolean	isLeaderControllerOfControllerCluster()	This function is used to detect whether current node is the leader controller of controller cluster.
boolean	isParent()	Check whether the controller works as a parent controller
boolean	isResourceStillAlive(String resourceName)	Test if a given helix resource is still alive (existent in ZK).
boolean	isRTTopicDeletionPermittedByAllControllers(String clusterName, String storeName)
boolean	isSSLEnabledForPush(String clusterName, String storeName)	Return whether ssl is enabled for the given store for push.
boolean	isSslToKafka()	Test if ssl is enabled to Kafka.
boolean	isStorageNodeNewerOrEqualTo(String clusterName, String instanceId, StorageNodeStatus oldStatus)	Compare the current storage node status and the given storage node status to check is the current one is "Newer" or "Equal" to the given one.
boolean	isStoreMigrationAllowed(String clusterName)	Test if the store migration is allowed for a cluster.
Pair<Boolean,String>	isStoreVersionReadyForDataRecovery(String clusterName, String storeName, int version, String sourceFabric, String destinationFabric, Optional<Integer> sourceAmplificationFactor)	Check if the store version's previous states and resources are cleaned up and ready to start data recovery.
boolean	isTopicTruncated(String kafkaTopicName)	Check if a kafka topic is absent or truncated.
boolean	isTopicTruncatedBasedOnRetention(long retention)	Test if retention is less than the configured DEPRECATED_TOPIC_MAX_RETENTION_MS value.
boolean	isTopicTruncatedBasedOnRetention(String kafkaTopicName, long retentionTime)	Topic should also be considered to get cleaned up if: retention is less than the configured ConfigKeys.DEPRECATED_TOPIC_MAX_RETENTION_MS value.
void	killOfflinePush(String clusterName, String kafkaTopic, boolean isForcedKill)	Kill an offline push if it ran into errors or the corresponding version is being retired.
Map<String,RegionPushDetails>	listStorePushInfo(String clusterName, String storeName, boolean isPartitionDetailEnabled)
protected static HybridStoreConfig	mergeNewSettingsIntoOldHybridStoreConfig(Store oldStore, Optional<Long> hybridRewindSeconds, Optional<Long> hybridOffsetLagThreshold, Optional<Long> hybridTimeLagThreshold, Optional<DataReplicationPolicy> hybridDataReplicationPolicy, Optional<BufferReplayPolicy> bufferReplayPolicy, Optional<String> realTimeTopicName)	Used by both the VeniceHelixAdmin and the VeniceParentHelixAdmin
void	migrateStore(String srcClusterName, String destClusterName, String storeName)	Main implementation for migrating a store from its source cluster to a new destination cluster.
Pair<NodeReplicasReadinessState,List<Replica>>	nodeReplicaReadiness(String cluster, String helixNodeId)	helixNodeId nodeId of helix participant.
Version	peekNextVersion(String clusterName, String storeName)
void	prepareDataRecovery(String clusterName, String storeName, int version, String sourceFabric, String destinationFabric, Optional<Integer> sourceAmplificationFactor)	Prepare for data recovery in the destination fabric.
DerivedSchemaEntry	removeDerivedSchema(String clusterName, String storeName, int valueSchemaId, int derivedSchemaId)	Remove an existing derived schema
void	removeInstanceFromAllowList(String clusterName, String helixNodeId)	Remove the given helix nodeId from the allowlist in ZK.
void	removeStorageNode(String clusterName, String instanceId)	Remove one storage node from the given cluster.
void	removeStoreFromGraveyard(String clusterName, String storeName)
void	replicateAddVersionAndStartIngestion(String clusterName, String storeName, String pushJobId, int versionNumber, int numberOfPartitions, Version.PushType pushType, String remoteKafkaBootstrapServers, long rewindTimeInSecondsOverride, int replicationMetadataVersionId)	This method is invoked in parent controllers to replicate new version signals for migrating store.
void	replicateUpdateStore(String clusterName, String storeName, UpdateStoreQueryParams params)	This method is invoked in parent controllers for store migration.
void	retireOldStoreVersions(String clusterName, String storeName, boolean deleteBackupOnStartPush, int currentVersionBeforePush)	For a given store, determine its versions to delete based on the BackupStrategy settings and execute the deletion in the cluster (including all its resources).
OfflinePushStatus	retrievePushStatus(String clusterName, StoreInfo store)
void	rollbackToBackupVersion(String clusterName, String storeName, String regionFilter)	Set backup version as current version in a child region.
void	rollForwardToFutureVersion(String clusterName, String storeName, String regionFilter)
void	sendHeartbeatToSystemStore(String clusterName, String storeName, long heartbeatTimeStamp)	Send a heartbeat timestamp to targeted system store.
void	sendKillMessageToParticipantStore(String clusterName, String kafkaTopic)
void	sendPushJobDetails(PushJobStatusRecordKey key, PushJobDetails value)	Lazy initialize a Venice writer for an internal real time topic store of push job details records.
void	setAdminConsumerService(String clusterName, AdminConsumerService service)
void	setPushJobDetailsStoreClient(AvroSpecificStoreClient<PushJobStatusRecordKey,PushJobDetails> client)
void	setStoreConfigForMigration(String storeName, String srcClusterName, String destClusterName)	Update "migrationDestCluster" and "migrationSrcCluster" fields of the "/storeConfigs/storeName" znode.
void	setStoreCurrentVersion(String clusterName, String storeName, int versionNumber)	Update the current version of a specified store.
void	setStoreLargestUsedVersion(String clusterName, String storeName, int versionNumber)	Update the largest used version number of a specified store.
void	setStoreOwner(String clusterName, String storeName, String owner)	Update the owner of a specified store.
void	setStorePartitionCount(String clusterName, String storeName, int partitionCount)	Since partition check/calculation only happens when adding new store version, setStorePartitionCount(String, String, int) would only change the number of partition for the following pushes.
void	setStorePushStrategyForMigration(String voldemortStoreName, String strategy)	Unsupported operation in the child controller.
void	setStoreReadability(String clusterName, String storeName, boolean desiredReadability)	Update the readability of a specified store.
void	setStoreReadWriteability(String clusterName, String storeName, boolean isAccessible)	Update both readability and writability of a specified store.
void	setStoreWriteability(String clusterName, String storeName, boolean desiredWriteability)	Update the writability of a specified store.
void	skipAdminMessage(String clusterName, long offset, boolean skipDIV)	The admin consumption task tries to deal with failures to process an admin message by retrying.
void	startInstanceMonitor(String clusterName)
void	stop(String clusterName)	Stop the helix controller for a single cluster.
void	stopMonitorOfflinePush(String clusterName, String topic, boolean deletePushStatus, boolean isForcedDelete)
void	stopVeniceController()	Stop the entire controller but not only the helix controller for a single cluster.
void	storeMetadataUpdate(String clusterName, String storeName, VeniceHelixAdmin.StoreMetadataOperation operation)	Update the store metadata by applying provided operation.
void	topicCleanupWhenPushComplete(String clusterName, String storeName, int versionNumber)	In this function, Controller will setup proper compaction strategy when the push job is full completed, and here are the reasons to set it up after the job completes: 1.
boolean	truncateKafkaTopic(String kafkaTopicName)	We don't actually truncate any Kafka topic here; we just update the retention time.
boolean	truncateKafkaTopic(String topicName, long retentionTime)	Truncate a Kafka topic by setting its retention time to the input value.
void	updateAclForStore(String clusterName, String storeName, String accessPermissions)	Provision a new set of ACL for a venice store and its associated kafka topic.
void	updateAdminTopicMetadata(String clusterName, long executionId, Optional<String> storeName, Optional<Long> offset, Optional<Long> upstreamOffset)	Update cluster-level execution id, offset and upstream offset.
void	updateClusterConfig(String clusterName, UpdateClusterConfigQueryParams params)	Update the LiveClusterConfig at runtime for a specified cluster.
void	updateClusterDiscovery(String storeName, String oldCluster, String newCluster, String initiatingCluster)	Update the cluster discovery of a given store by writing to the StoreConfig ZNode.
void	updateRoutersClusterConfig(String clusterName, Optional<Boolean> isThrottlingEnable, Optional<Boolean> isQuotaRebalancedEnable, Optional<Boolean> isMaxCapacityProtectionEnabled, Optional<Integer> expectedRouterCount)	Update the cluster level for all routers.
void	updateStoragePersona(String clusterName, String name, UpdateStoragePersonaQueryParams queryParams)
void	updateStore(String clusterName, String storeName, UpdateStoreQueryParams params)	TODO: some logics are in parent controller VeniceParentHelixAdmin #updateStore and some are in the child controller here.
void	validateAndMaybeRetrySystemStoreAutoCreation(String clusterName, String storeName, VeniceSystemStoreType systemStoreType)	Check the creation results of a user store's system store.
List<Version>	versionsForStore(String clusterName, String storeName)
boolean	whetherEnableBatchPushFromAdmin(String storeName)	Test if a store is allowed for a batch push.
void	wipeCluster(String clusterName, String fabric, Optional<String> storeName, Optional<Integer> versionNum)	Delete stores from the cluster including both store data and metadata.
void	writeEndOfPush(String clusterName, String storeName, int versionNumber, boolean alsoWriteStartOfPush)	Create a local Venice writer based on store version info and, for each partition, use the writer to send END_OF_PUSH and END_OF_SEGMENT control messages to Kafka.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Methods inherited from interface com.linkedin.venice.controller.Admin

addValueSchema, createStore, createStore, getDatacenterCount, hasWritePermissionToBatchJobHeartbeatStore, incrementVersionIdempotent, incrementVersionIdempotent

Field Details

INTERNAL_STORE_GET_RRT_TOPIC_ATTEMPTS

protected static final int INTERNAL_STORE_GET_RRT_TOPIC_ATTEMPTS

See Also:

• Constant Field Values

INTERNAL_STORE_RTT_RETRY_BACKOFF_MS

protected static final long INTERNAL_STORE_RTT_RETRY_BACKOFF_MS

pubSubTopicRepository

protected final PubSubTopicRepository pubSubTopicRepository

Constructor Details

VeniceHelixAdmin

public VeniceHelixAdmin(VeniceControllerMultiClusterConfig multiClusterConfigs,
 io.tehuti.metrics.MetricsRepository metricsRepository,
 com.linkedin.d2.balancer.D2Client d2Client,
 PubSubTopicRepository pubSubTopicRepository,
 PubSubClientsFactory pubSubClientsFactory)

VeniceHelixAdmin

public VeniceHelixAdmin(VeniceControllerMultiClusterConfig multiClusterConfigs,
 io.tehuti.metrics.MetricsRepository metricsRepository,
 boolean sslEnabled,
 @Nonnull
 com.linkedin.d2.balancer.D2Client d2Client,
 Optional<SSLConfig> sslConfig,
 Optional<DynamicAccessController> accessController,
 Optional<ICProvider> icProvider,
 PubSubTopicRepository pubSubTopicRepository,
 PubSubClientsFactory pubSubClientsFactory,
 List<ClusterLeaderInitializationRoutine> additionalInitRoutines)

Method Details

startInstanceMonitor

public void startInstanceMonitor(String clusterName)

Specified by:

startInstanceMonitor in interface Admin

getLiveInstanceMonitor

public LiveInstanceMonitor getLiveInstanceMonitor(String clusterName)

clearInstanceMonitor

public void clearInstanceMonitor(String clusterName)

Specified by:

clearInstanceMonitor in interface Admin

getZkClient

public org.apache.helix.zookeeper.impl.client.ZkClient getZkClient()

getExecutionIdAccessor

public ExecutionIdAccessor getExecutionIdAccessor()

getAdapterSerializer

public HelixAdapterSerializer getAdapterSerializer()

initStorageCluster

public void initStorageCluster(String clusterName)

Create and configure the Venice storage cluster with required properties in Helix and waits the resource's (partial) partition to appear in the external view.

Specified by:

initStorageCluster in interface Admin

Parameters:

clusterName - Venice cluster name.

isResourceStillAlive

public boolean isResourceStillAlive(String resourceName)

Test if a given helix resource is still alive (existent in ZK).

Specified by:

isResourceStillAlive in interface Admin

Parameters:

resourceName - Helix resource name.

Returns:

true if resource exists; false otherwise.

isClusterValid

public boolean isClusterValid(String clusterName)

Test if a cluster is valid (in Helix cluster list).

Specified by:

isClusterValid in interface Admin

Parameters:

clusterName - Venice cluster name.

Returns:

true if input cluster is in Helix cluster list; false otherwise.

getHelixAdmin

protected org.apache.helix.HelixAdmin getHelixAdmin()

createStore

public void createStore(String clusterName,
 String storeName,
 String owner,
 String keySchema,
 String valueSchema,
 boolean isSystemStore,
 Optional<String> accessPermissions)

Create a new ZK store and its configuration in the store repository and create schemas in the schema repository.

Specified by:

createStore in interface Admin

Parameters:

clusterName - Venice cluster where the store locates.

storeName - name of the store.

owner - owner of the store.

keySchema - key schema of the store.

valueSchema - value schema of the store.

isSystemStore - if the store is a system store.

accessPermissions - json string representing the access-permissions.

deleteStore

public void deleteStore(String clusterName,
 String storeName,
 int largestUsedVersionNumber,
 boolean waitOnRTTopicDeletion)

This method will delete store data, metadata, version and rt topics One exception is for stores with isMigrating flag set. In that case, the corresponding kafka topics and storeConfig will not be deleted so that they are still available for the cloned store.

Specified by:

deleteStore in interface Admin

sendPushJobDetails

public void sendPushJobDetails(PushJobStatusRecordKey key,
 PushJobDetails value)

Lazy initialize a Venice writer for an internal real time topic store of push job details records. Use this writer to put a pair of push job detail record (key and value).

Specified by:

sendPushJobDetails in interface Admin

Parameters:

key - key with which the specified value is to be associated.

value - value to be associated with the specified key.

getPushJobDetails

public PushJobDetails getPushJobDetails(@Nonnull
 PushJobStatusRecordKey key)

Specified by:

getPushJobDetails in interface Admin

Returns:

the value to which the specified key is mapped from the Venice internal real time topic store.

getBatchJobHeartbeatValue

public BatchJobHeartbeatValue getBatchJobHeartbeatValue(@Nonnull
 BatchJobHeartbeatKey batchJobHeartbeatKey)

Specified by:

getBatchJobHeartbeatValue in interface Admin

Returns:

the value to which the specified key is mapped from the Venice internal BATCH_JOB_HEARTBEAT_STORE topic store.

writeEndOfPush

public void writeEndOfPush(String clusterName,
 String storeName,
 int versionNumber,
 boolean alsoWriteStartOfPush)

Create a local Venice writer based on store version info and, for each partition, use the writer to send END_OF_PUSH and END_OF_SEGMENT control messages to Kafka.

Specified by:

writeEndOfPush in interface Admin

Parameters:

clusterName - name of the Venice cluster.

storeName - name of the store.

versionNumber - store version number.

alsoWriteStartOfPush - if Venice writer sends a START_OF_PUSH control message first.

whetherEnableBatchPushFromAdmin

public boolean whetherEnableBatchPushFromAdmin(String storeName)

Test if a store is allowed for a batch push.

Specified by:

whetherEnableBatchPushFromAdmin in interface Admin

Parameters:

storeName - name of a store.

Returns:

true is the store is a participant system store or if Venice is running in single-region mode

isStoreMigrationAllowed

public boolean isStoreMigrationAllowed(String clusterName)

Test if the store migration is allowed for a cluster. It reads the value "allow.store.migration" from the "/clusterName/ClusterConfig" znode.

Specified by:

isStoreMigrationAllowed in interface Admin

Parameters:

clusterName - name of Venice cluster.

Returns:

true if store migration is allowed for the input cluster; false otherwise.

migrateStore

public void migrateStore(String srcClusterName,
 String destClusterName,
 String storeName)

Main implementation for migrating a store from its source cluster to a new destination cluster. A new store (with same properties, e.g. name, owner, key schema, value schema) is created at the destination cluster and its StoreInfo is also cloned. For a store with enabled meta system store or enabled davinci push status, those system stores are also migrated. Different store versions are evaluated for the migration. For those versions to be migrated, it triggers the ADD_VERSION and starts ingestion at the destination cluster.

Specified by:

migrateStore in interface Admin

Parameters:

srcClusterName - name of the source cluster.

destClusterName - name of the destination cluster.

storeName - name of the target store.

clearIngestionKillMessageAndVerify

public void clearIngestionKillMessageAndVerify(String clusterName,
 String versionTopicName)

Clear KILL messages from a participant system store.

getControllerClientMap

public Map<String,ControllerClient> getControllerClientMap(String clusterName)

completeMigration

public void completeMigration(String srcClusterName,
 String destClusterName,
 String storeName)

Specified by:

completeMigration in interface Admin

See Also:

• updateClusterDiscovery(String, String, String, String)

abortMigration

public void abortMigration(String srcClusterName,
 String destClusterName,
 String storeName)

Abort store migration by resetting migration flag at the source cluster, resetting storeConfig, and updating "cluster" in "/storeConfigs" znode back to the source cluster.

Specified by:

abortMigration in interface Admin

Parameters:

srcClusterName - name of the source cluster.

destClusterName - name of the destination cluster.

storeName - name of the store in migration.

updateClusterDiscovery

public void updateClusterDiscovery(String storeName,
 String oldCluster,
 String newCluster,
 String initiatingCluster)

Description copied from interface: Admin

Update the cluster discovery of a given store by writing to the StoreConfig ZNode.

Specified by:

updateClusterDiscovery in interface Admin

Parameters:

storeName - of the store.

oldCluster - for the store.

newCluster - for the store.

initiatingCluster - that is making the update. This is needed because in the case of store migration sometimes the update is not made by the leader of the current cluster but instead the leader of the source cluster.

See Also:

• Admin.updateClusterDiscovery(String, String, String, String)

checkPreConditionForCreateStore

protected void checkPreConditionForCreateStore(String clusterName,
 String storeName,
 String keySchema,
 String valueSchema,
 boolean allowSystemStore,
 boolean skipLingeringResourceCheck)

Check whether Controller should block the incoming store creation. Inside this function, there is a logic to check whether there are any lingering resources since the requested store could be just deleted recently. This check should be skipped in Child Controller, but only enabled in Parent Controller because of the following reasons: 1. Parent Controller has the strict order that the system store must be created before the host Venice store. 2. Child Controller doesn't have this strict order since the admin messages of Child Controller could be executed in parallel since they are different store names. So when such kind of race condition happens, it will cause a dead loop: a. The version creation of system store will create a RT topic in Parent Cluster. b. The RT topic will be mirrored by KMM to the Child Cluster. c. The version creation admin message of system store will be blocked in Child Controller since the host Venice store doesn't exist. d. The store creation admin message of the host Venice store will be blocked in Child Controller because of lingering resource check (RT topic of its system store already exists, which is created by KMM). TODO: Evaluate if this code needs to change now that KMM has been deprecated. In the future, once Venice gets rid of KMM, the topic won't be automatically created by KMM, and this race condition will be addressed. So far, Child Controller will skip lingering resource check when handling store creation admin message.

addVersionAndStartIngestion

public void addVersionAndStartIngestion(String clusterName,
 String storeName,
 String pushJobId,
 int versionNumber,
 int numberOfPartitions,
 Version.PushType pushType,
 String remoteKafkaBootstrapServers,
 long rewindTimeInSecondsOverride,
 int replicationMetadataVersionId,
 boolean versionSwapDeferred,
 int repushSourceVersion)

Description copied from interface: Admin

This method behaves differently in VeniceHelixAdmin and VeniceParentHelixAdmin.

Specified by:

addVersionAndStartIngestion in interface Admin

addVersionAndStartIngestion

public void addVersionAndStartIngestion(String clusterName,
 String storeName,
 String pushJobId,
 int versionNumber,
 int numberOfPartitions,
 Version.PushType pushType,
 String remoteKafkaBootstrapServers,
 long rewindTimeInSecondsOverride,
 int replicationMetadataVersionId,
 boolean versionSwapDeferred,
 String targetedRegions,
 int repushSourceVersion)

This is a wrapper for VeniceHelixAdmin#addVersion but performs additional operations needed for add version invoked from the admin channel. Therefore, this method is mainly invoked from the admin task upon processing an add version message.

replicateAddVersionAndStartIngestion

public void replicateAddVersionAndStartIngestion(String clusterName,
 String storeName,
 String pushJobId,
 int versionNumber,
 int numberOfPartitions,
 Version.PushType pushType,
 String remoteKafkaBootstrapServers,
 long rewindTimeInSecondsOverride,
 int replicationMetadataVersionId)

This method is invoked in parent controllers to replicate new version signals for migrating store.

addVersionAndTopicOnly

public Pair<Boolean,Version> addVersionAndTopicOnly(String clusterName,
 String storeName,
 String pushJobId,
 int versionNumber,
 int numberOfPartitions,
 int replicationFactor,
 boolean sendStartOfPush,
 boolean sorted,
 Version.PushType pushType,
 String compressionDictionary,
 String remoteKafkaBootstrapServers,
 Optional<String> sourceGridFabric,
 long rewindTimeInSecondsOverride,
 int replicationMetadataVersionId,
 Optional<String> emergencySourceRegion,
 boolean versionSwapDeferred)

addVersionAndTopicOnly

public Pair<Boolean,Version> addVersionAndTopicOnly(String clusterName,
 String storeName,
 String pushJobId,
 int versionNumber,
 int numberOfPartitions,
 int replicationFactor,
 boolean sendStartOfPush,
 boolean sorted,
 Version.PushType pushType,
 String compressionDictionary,
 String remoteKafkaBootstrapServers,
 Optional<String> sourceGridFabric,
 long rewindTimeInSecondsOverride,
 int replicationMetadataVersionId,
 Optional<String> emergencySourceRegion,
 boolean versionSwapDeferred,
 String targetedRegions,
 int repushSourceVersion)

A wrapper to invoke VeniceHelixAdmin#addVersion to only increment the version and create the topic(s) needed without starting ingestion.

addVersionOnly

public Version addVersionOnly(String clusterName,
 String storeName,
 String pushJobId,
 int versionNumber,
 int numberOfPartitions,
 Version.PushType pushType,
 String remoteKafkaBootstrapServers,
 long rewindTimeInSecondsOverride,
 int replicationMetadataVersionId)

Only add version to the store without creating the topic or start ingestion. Used to sync version metadata in the parent fabric during store migration.

addSpecificVersion

public boolean addSpecificVersion(String clusterName,
 String storeName,
 Version version)

TODO refactor addVersion to these broken down methods instead of doing everything in one giant method. Perform add version to a given store with the provided Version

createSpecificVersionTopic

public void createSpecificVersionTopic(String clusterName,
 String storeName,
 Version version)

Create the corresponding version topic based on the provided Version

createHelixResourceAndStartMonitoring

public void createHelixResourceAndStartMonitoring(String clusterName,
 String storeName,
 Version version)

Create Helix-resources for a given storage node cluster and starts monitoring a new push.

incrementVersionIdempotent

public Version incrementVersionIdempotent(String clusterName,
 String storeName,
 String pushJobId,
 int numberOfPartitions,
 int replicationFactor,
 Version.PushType pushType,
 boolean sendStartOfPush,
 boolean sorted,
 String compressionDictionary,
 Optional<String> sourceGridFabric,
 Optional<X509Certificate> requesterCert,
 long rewindTimeInSecondsOverride,
 Optional<String> emergencySourceRegion,
 boolean versionSwapDeferred,
 String targetedRegions,
 int repushSourceVersion)

Note: this currently use the pushID to guarantee idempotence, unexpected behavior may result if multiple batch jobs push to the same store at the same time.

Specified by:

incrementVersionIdempotent in interface Admin

getStartedVersion

protected static Optional<Version> getStartedVersion(Store store)

The intended semantic is to use this method to find the version that something is currently pushing to. It looks at all versions greater than the current version and identifies the version with a status of STARTED. If there is no STARTED version, it creates a new one for the push to use. This means we cannot use this method to support multiple concurrent pushes.

Parameters:

store -

Returns:

the started version if there is only one, throws an exception if there is an error version with a greater number than the current version. Otherwise returns Optional.empty()

getReplicationMetadataSchema

public Optional<org.apache.avro.Schema> getReplicationMetadataSchema(String clusterName,
 String storeName,
 int valueSchemaID,
 int rmdVersionID)

Specified by:

getReplicationMetadataSchema in interface Admin

Returns:

replication metadata schema for a store in a cluster with specified schema ID and RMD protocol version ID.

getReferenceVersionForStreamingWrites

public Version getReferenceVersionForStreamingWrites(String clusterName,
 String storeName,
 String pushJobId)

Specified by:

getReferenceVersionForStreamingWrites in interface Admin

getIncrementalPushVersion

public Version getIncrementalPushVersion(String clusterName,
 String storeName,
 String pushJobId)

Specified by:

getIncrementalPushVersion in interface Admin

getCurrentVersion

public int getCurrentVersion(String clusterName,
 String storeName)

Specified by:

getCurrentVersion in interface Admin

Returns:

The current version number of an input store in the specified Venice cluster or Store.NON_EXISTING_VERSION if none exists.

getFutureVersion

public int getFutureVersion(String clusterName,
 String storeName)

Specified by:

getFutureVersion in interface Admin

Returns:

Returns the online (completed, but not yet swapped) or future version with ongoing ingestion else if none exists returns Store.NON_EXISTING_VERSION

getBackupVersion

public int getBackupVersion(String clusterName,
 String storeName)

Specified by:

getBackupVersion in interface Admin

getOnlineFutureVersion

public int getOnlineFutureVersion(String clusterName,
 String storeName)

getCurrentVersionsForMultiColos

public Map<String,Integer> getCurrentVersionsForMultiColos(String clusterName,
 String storeName)

Specified by:

getCurrentVersionsForMultiColos in interface Admin

getRepushInfo

public RepushInfo getRepushInfo(String clusterName,
 String storeName,
 Optional<String> fabricName)

Specified by:

getRepushInfo in interface Admin

Returns:

a new RepushInfo object with specified store info.

getFutureVersionsForMultiColos

public Map<String,String> getFutureVersionsForMultiColos(String clusterName,
 String storeName)

Specified by:

getFutureVersionsForMultiColos in interface Admin

See Also:

• Admin.getFutureVersionsForMultiColos(String, String)

getBackupVersionsForMultiColos

public Map<String,String> getBackupVersionsForMultiColos(String clusterName,
 String storeName)

Specified by:

getBackupVersionsForMultiColos in interface Admin

peekNextVersion

public Version peekNextVersion(String clusterName,
 String storeName)

Specified by:

peekNextVersion in interface Admin

Returns:

the next version without adding the new version to the store.

deleteAllVersionsInStore

public List<Version> deleteAllVersionsInStore(String clusterName,
 String storeName)

Description copied from interface: Admin

Delete all of venice versions in given store(including venice resource, kafka topic, offline pushs and all related resources).

Specified by:

deleteAllVersionsInStore in interface Admin

See Also:

• Admin.deleteAllVersionsInStore(String, String)

deleteOldVersionInStore

public void deleteOldVersionInStore(String clusterName,
 String storeName,
 int versionNum)

Description copied from interface: Admin

Delete the given version from the store. If the given version is the current version, an exception will be thrown.

Specified by:

deleteOldVersionInStore in interface Admin

See Also:

• Admin.deleteOldVersionInStore(String, String, int)

deleteOneStoreVersion

public void deleteOneStoreVersion(String clusterName,
 String storeName,
 int versionNumber)

Delete version from cluster, removing all related resources

Specified by:

deleteOneStoreVersion in interface StoreCleaner

isRTTopicDeletionPermittedByAllControllers

public boolean isRTTopicDeletionPermittedByAllControllers(String clusterName,
 String storeName)

Specified by:

isRTTopicDeletionPermittedByAllControllers in interface Admin

retireOldStoreVersions

public void retireOldStoreVersions(String clusterName,
 String storeName,
 boolean deleteBackupOnStartPush,
 int currentVersionBeforePush)

For a given store, determine its versions to delete based on the BackupStrategy settings and execute the deletion in the cluster (including all its resources). It also truncates Kafka topics and Helix resources.

Specified by:

retireOldStoreVersions in interface StoreCleaner

Parameters:

clusterName - name of a cluster.

storeName - name of the store to retire.

deleteBackupOnStartPush - indicate if it is called in a start-of-push workflow.

currentVersionBeforePush - current version before a new push.

topicCleanupWhenPushComplete

public void topicCleanupWhenPushComplete(String clusterName,
 String storeName,
 int versionNumber)

In this function, Controller will setup proper compaction strategy when the push job is full completed, and here are the reasons to set it up after the job completes: 1. For batch push jobs to batch-only store, there is no impact. There could still be duplicate entries because of speculative executions in map-reduce job, but we are not planning to clean them up now. 2. For batch push jobs to hybrid/incremental stores, if the compaction is enabled at the beginning of the job, Kafka compaction could kick in during push job, and storage node could detect DIV error, such as missing messages, checksum mismatch, because speculative execution could produce duplicate entries, and we don't want to fail the push in this scenario and we still want to perform the strong DIV validation in batch push, so we could only enable compaction after the batch push completes. 3. For GF jobs to hybrid store, it is similar as #2, and it contains duplicate entries because there is no de-dedup happening anywhere. With this way, when load rebalance happens for hybrid/incremental stores, DIV error could be detected during ingestion at any phase since compaction might be enabled long-time ago. So in storage node, we need to add one more safeguard before throwing the DIV exception to check whether the topic is compaction-enabled or not. Since Venice is not going to change the compaction policy between non-compact and compact back and forth, checking whether topic is compaction-enabled or not when encountering DIV error should be good enough.

Specified by:

topicCleanupWhenPushComplete in interface StoreCleaner

isTopicTruncated

public boolean isTopicTruncated(String kafkaTopicName)

Check if a kafka topic is absent or truncated.

Specified by:

isTopicTruncated in interface Admin

See Also:

• ConfigKeys.DEPRECATED_TOPIC_MAX_RETENTION_MS

isTopicTruncatedBasedOnRetention

public boolean isTopicTruncatedBasedOnRetention(long retention)

Test if retention is less than the configured DEPRECATED_TOPIC_MAX_RETENTION_MS value.

Specified by:

isTopicTruncatedBasedOnRetention in interface Admin

Returns:

true if the specified retention is below the configuration; false otherwise.

See Also:

• ConfigKeys.DEPRECATED_TOPIC_MAX_RETENTION_MS

isTopicTruncatedBasedOnRetention

public boolean isTopicTruncatedBasedOnRetention(String kafkaTopicName,
 long retentionTime)

Topic should also be considered to get cleaned up if: retention is less than the configured ConfigKeys.DEPRECATED_TOPIC_MAX_RETENTION_MS value. or 1. Topic retention equals fatalDataValidationFailureRetentionMs. 2. Topic is a version topic. 3. fatalDataValidationFailureRetentionMs has already passed since its creation.

Specified by:

isTopicTruncatedBasedOnRetention in interface Admin

getMinNumberOfUnusedKafkaTopicsToPreserve

public int getMinNumberOfUnusedKafkaTopicsToPreserve()

Specified by:

getMinNumberOfUnusedKafkaTopicsToPreserve in interface Admin

Returns:

the controller configuration value for MIN_NUMBER_OF_UNUSED_KAFKA_TOPICS_TO_PRESERVE.

See Also:

• ConfigKeys.MIN_NUMBER_OF_UNUSED_KAFKA_TOPICS_TO_PRESERVE

truncateKafkaTopic

public boolean truncateKafkaTopic(String kafkaTopicName)

We don't actually truncate any Kafka topic here; we just update the retention time.

Specified by:

truncateKafkaTopic in interface Admin

Parameters:

kafkaTopicName -

Returns:

truncateKafkaTopic

public boolean truncateKafkaTopic(String topicName,
 long retentionTime)

Description copied from interface: Admin

Truncate a Kafka topic by setting its retention time to the input value.

Specified by:

truncateKafkaTopic in interface Admin

Parameters:

topicName - the name of the topic to truncate.

retentionTime - the retention time in milliseconds to set for the topic.

Returns:

true if truncating this topic successfully. false otherwise.

See Also:

• Admin.truncateKafkaTopic(String, long)

versionsForStore

public List<Version> versionsForStore(String clusterName,
 String storeName)

Specified by:

versionsForStore in interface Admin

Returns:

all versions of the specified store from a cluster.

getAllStores

public List<Store> getAllStores(String clusterName)

Specified by:

getAllStores in interface Admin

Returns:

all stores in the specified cluster.

getAllStoreStatuses

public Map<String,String> getAllStoreStatuses(String clusterName)

Description copied from interface: Admin

Get the statuses of all stores. The store status is decided by the current version. For example, if one partition only have 2 ONLINE replicas in the current version, we say this store is under replicated. Refer to StoreStatus for the definition of each status.

Specified by:

getAllStoreStatuses in interface Admin

Returns:

a map whose key is store name and value is store's status.

See Also:

• Admin.getAllStoreStatuses(String)

hasStore

public boolean hasStore(String clusterName,
 String storeName)

Test if the input store exists in a cluster.

Specified by:

hasStore in interface Admin

Parameters:

clusterName - name of a cluster.

storeName - name of a store.

Returns:

ture if store exists in the cluster. false otherwise.

getStore

public Store getStore(String clusterName,
 String storeName)

Specified by:

getStore in interface Admin

Returns:

Store object reference from the input store name.

setStoreCurrentVersion

public void setStoreCurrentVersion(String clusterName,
 String storeName,
 int versionNumber)

Update the current version of a specified store.

Specified by:

setStoreCurrentVersion in interface Admin

rollForwardToFutureVersion

public void rollForwardToFutureVersion(String clusterName,
 String storeName,
 String regionFilter)

Specified by:

rollForwardToFutureVersion in interface Admin

rollbackToBackupVersion

public void rollbackToBackupVersion(String clusterName,
 String storeName,
 String regionFilter)

Set backup version as current version in a child region.

Specified by:

rollbackToBackupVersion in interface Admin

getBackupVersionNumber

public int getBackupVersionNumber(List<Version> versions,
 int currentVersion)

Get backup version number, the largest online version number that is less than the current version number

setStoreLargestUsedVersion

public void setStoreLargestUsedVersion(String clusterName,
 String storeName,
 int versionNumber)

Update the largest used version number of a specified store.

Specified by:

setStoreLargestUsedVersion in interface Admin

setStoreOwner

public void setStoreOwner(String clusterName,
 String storeName,
 String owner)

Update the owner of a specified store.

Specified by:

setStoreOwner in interface Admin

setStorePartitionCount

public void setStorePartitionCount(String clusterName,
 String storeName,
 int partitionCount)

Since partition check/calculation only happens when adding new store version, setStorePartitionCount(String, String, int) would only change the number of partition for the following pushes. Current version would not be changed.

Specified by:

setStorePartitionCount in interface Admin

setStoreWriteability

public void setStoreWriteability(String clusterName,
 String storeName,
 boolean desiredWriteability)

Update the writability of a specified store.

Specified by:

setStoreWriteability in interface Admin

setStoreReadability

public void setStoreReadability(String clusterName,
 String storeName,
 boolean desiredReadability)

Update the readability of a specified store.

Specified by:

setStoreReadability in interface Admin

setStoreReadWriteability

public void setStoreReadWriteability(String clusterName,
 String storeName,
 boolean isAccessible)

Update both readability and writability of a specified store.

Specified by:

setStoreReadWriteability in interface Admin

getMetaStoreValue

public StoreMetaValue getMetaStoreValue(StoreMetaKey metaKey,
 String storeName)

Specified by:

getMetaStoreValue in interface Admin

getInUseValueSchemaIds

public Set<Integer> getInUseValueSchemaIds(String clusterName,
 String storeName)

Specified by:

getInUseValueSchemaIds in interface Admin

deleteValueSchemas

public void deleteValueSchemas(String clusterName,
 String storeName,
 Set<Integer> unusedValueSchemaIds)

Description copied from interface: Admin

Deletes a store's values schema with ids `except` the ids passed in the argument inuseValueSchemaIds

Specified by:

deleteValueSchemas in interface Admin

updateStore

public void updateStore(String clusterName,
 String storeName,
 UpdateStoreQueryParams params)

TODO: some logics are in parent controller VeniceParentHelixAdmin #updateStore and some are in the child controller here. Need to unify them in the future.

Specified by:

updateStore in interface Admin

updateClusterConfig

public void updateClusterConfig(String clusterName,
 UpdateClusterConfigQueryParams params)

Update the LiveClusterConfig at runtime for a specified cluster.

Specified by:

updateClusterConfig in interface Admin

Parameters:

clusterName - name of the Venice cluster.

params - parameters to update.

replicateUpdateStore

public void replicateUpdateStore(String clusterName,
 String storeName,
 UpdateStoreQueryParams params)

This method is invoked in parent controllers for store migration.

mergeNewSettingsIntoOldHybridStoreConfig

protected static HybridStoreConfig mergeNewSettingsIntoOldHybridStoreConfig(Store oldStore,
 Optional<Long> hybridRewindSeconds,
 Optional<Long> hybridOffsetLagThreshold,
 Optional<Long> hybridTimeLagThreshold,
 Optional<DataReplicationPolicy> hybridDataReplicationPolicy,
 Optional<BufferReplayPolicy> bufferReplayPolicy,
 Optional<String> realTimeTopicName)

Used by both the VeniceHelixAdmin and the VeniceParentHelixAdmin

Parameters:

oldStore - Existing Store that is the source for updates. This object will not be modified by this method.

hybridRewindSeconds - Optional is present if the returned object should include a new rewind time

hybridOffsetLagThreshold - Optional is present if the returned object should include a new offset lag threshold

Returns:

null if oldStore has no hybrid configs and optionals are not present, otherwise a fully specified HybridStoreConfig

storeMetadataUpdate

public void storeMetadataUpdate(String clusterName,
 String storeName,
 VeniceHelixAdmin.StoreMetadataOperation operation)

Update the store metadata by applying provided operation.

Parameters:

clusterName - name of the cluster.

storeName - name of the to be updated store.

operation - the defined operation that update the store.

getStorageEngineOverheadRatio

public double getStorageEngineOverheadRatio(String clusterName)

Specified by:

getStorageEngineOverheadRatio in interface Admin

Returns:

the configuration value for ConfigKeys.STORAGE_ENGINE_OVERHEAD_RATIO

containsHelixResource

public boolean containsHelixResource(String clusterName,
 String kafkaTopic)

Description copied from interface: StoreCleaner

This purpose of the function is to check if the given resource exists in the Helix cluster.

Specified by:

containsHelixResource in interface StoreCleaner

Parameters:

clusterName - The Venice cluster that the resource belongs to.

kafkaTopic - it's usually the store version name (version topic name).

Returns:

deleteHelixResource

public void deleteHelixResource(String clusterName,
 String kafkaTopic)

Description copied from interface: StoreCleaner

This purpose of the function is to delete the given resource from the Helix cluster. Different from StoreCleaner.deleteOneStoreVersion(String, String, int), this function will not check whether the store version is still a valid version inside Venice backend, and it will send the delete request to Helix cluster directly. Do enough sanity check before calling this function.

Specified by:

deleteHelixResource in interface StoreCleaner

Parameters:

clusterName - The Venice cluster that the resource belongs to.

kafkaTopic - It's usually the store version name (version topic name).

enableDisabledPartition

public void enableDisabledPartition(String clusterName,
 String kafkaTopic,
 boolean enableAll)

getKeySchema

public SchemaEntry getKeySchema(String clusterName,
 String storeName)

Specified by:

getKeySchema in interface Admin

Returns:

the key schema for the specified store.

getValueSchemas

public Collection<SchemaEntry> getValueSchemas(String clusterName,
 String storeName)

Specified by:

getValueSchemas in interface Admin

Returns:

the value schema for the specified store.

getDerivedSchemas

public Collection<DerivedSchemaEntry> getDerivedSchemas(String clusterName,
 String storeName)

Specified by:

getDerivedSchemas in interface Admin

Returns:

the derived schema for the specified store.

getValueSchemaId

public int getValueSchemaId(String clusterName,
 String storeName,
 String valueSchemaStr)

Specified by:

getValueSchemaId in interface Admin

Returns:

the schema id for the specified store and value schema.

getDerivedSchemaId

public GeneratedSchemaID getDerivedSchemaId(String clusterName,
 String storeName,
 String schemaStr)

Specified by:

getDerivedSchemaId in interface Admin

Returns:

the derived schema id for the specified store and derived schema.

getValueSchema

public SchemaEntry getValueSchema(String clusterName,
 String storeName,
 int id)

Specified by:

getValueSchema in interface Admin

Returns:

the derived schema for the specified store and id.

addValueSchema

public SchemaEntry addValueSchema(String clusterName,
 String storeName,
 String valueSchemaStr,
 DirectionalSchemaCompatibilityType expectedCompatibilityType)

Specified by:

addValueSchema in interface Admin

See Also:

• addValueSchema(String, String, String, int, DirectionalSchemaCompatibilityType)

addValueSchema

public SchemaEntry addValueSchema(String clusterName,
 String storeName,
 String valueSchemaStr,
 int schemaId,
 DirectionalSchemaCompatibilityType compatibilityType)

Add a new value schema for the given store with all specified properties and return a new SchemaEntry object containing the schema and its id.

Specified by:

addValueSchema in interface Admin

Returns:

an SchemaEntry object composed of a schema and its corresponding id.

addDerivedSchema

public DerivedSchemaEntry addDerivedSchema(String clusterName,
 String storeName,
 int valueSchemaId,
 String derivedSchemaStr)

Add a new derived schema for the given store with all specified properties and return a new DerivedSchemaEntry object containing the schema and its id.

Specified by:

addDerivedSchema in interface Admin

Returns:

an DerivedSchemaEntry object composed of specified properties.

addDerivedSchema

public DerivedSchemaEntry addDerivedSchema(String clusterName,
 String storeName,
 int valueSchemaId,
 int derivedSchemaId,
 String derivedSchemaStr)

Add a new derived schema for the given store with all specified properties.

Specified by:

addDerivedSchema in interface Admin

Returns:

an DerivedSchemaEntry object composed of specified properties.

removeDerivedSchema

public DerivedSchemaEntry removeDerivedSchema(String clusterName,
 String storeName,
 int valueSchemaId,
 int derivedSchemaId)

Description copied from interface: Admin

Remove an existing derived schema

Specified by:

removeDerivedSchema in interface Admin

Returns:

the derived schema that is deleted or null if the schema doesn't exist

See Also:

• Admin.removeDerivedSchema(String, String, int, int)

addSupersetSchema

public SchemaEntry addSupersetSchema(String clusterName,
 String storeName,
 String valueSchema,
 int valueSchemaId,
 String supersetSchemaStr,
 int supersetSchemaId)

Add a new superset schema for the given store with all specified properties.

Generate the superset schema off the current schema and latest superset schema (if any, if not pick the latest value schema) existing in the store. If the newly generated superset schema is unique add it to the store and update latestSuperSetValueSchemaId of the store.

Specified by:

addSupersetSchema in interface Admin

getReplicationMetadataSchemas

public Collection<RmdSchemaEntry> getReplicationMetadataSchemas(String clusterName,
 String storeName)

Specified by:

getReplicationMetadataSchemas in interface Admin

Returns:

a collection of ReplicationMetadataSchemaEntry object for the given store and cluster.

addReplicationMetadataSchema

public RmdSchemaEntry addReplicationMetadataSchema(String clusterName,
 String storeName,
 int valueSchemaId,
 int replicationMetadataVersionId,
 String replicationMetadataSchemaStr)

Create a new ReplicationMetadataSchemaEntry object with the given properties and add it into schema repository if no duplication.

Specified by:

addReplicationMetadataSchema in interface Admin

Returns:

ReplicationMetadataSchemaEntry object reference.

validateAndMaybeRetrySystemStoreAutoCreation

public void validateAndMaybeRetrySystemStoreAutoCreation(String clusterName,
 String storeName,
 VeniceSystemStoreType systemStoreType)

Check the creation results of a user store's system store. If the system store's current version is in error state, re-issue a new empty push and waits for the empty push to complete.

Specified by:

validateAndMaybeRetrySystemStoreAutoCreation in interface Admin

getStorageNodes

public List<String> getStorageNodes(String clusterName)

Specified by:

getStorageNodes in interface Admin

Returns:

a list of storage node instance names for a given cluster.

getHelixAdminClient

public HelixAdminClient getHelixAdminClient()

getDisabledPartitionStats

public DisabledPartitionStats getDisabledPartitionStats(String clusterName)

getStorageNodesStatus

public Map<String,String> getStorageNodesStatus(String clusterName,
 boolean enableReplica)

Specified by:

getStorageNodesStatus in interface Admin

Returns:

a map containing the storage node name and its connectivity status (InstanceStatus).

removeStorageNode

public void removeStorageNode(String clusterName,
 String instanceId)

Remove one storage node from the given cluster.

It removes the given helix nodeId from the allowlist in ZK and its associated resource in Helix.

Specified by:

removeStorageNode in interface Admin

stop

public void stop(String clusterName)

Description copied from interface: Admin

Stop the helix controller for a single cluster.

Specified by:

stop in interface Admin

See Also:

• Admin.stop(String)

stopVeniceController

public void stopVeniceController()

Description copied from interface: Admin

Stop the entire controller but not only the helix controller for a single cluster.

Specified by:

stopVeniceController in interface Admin

See Also:

• Admin.stopVeniceController()

getOffLinePushStatus

public Admin.OfflinePushStatusInfo getOffLinePushStatus(String clusterName,
 String kafkaTopic)

Description copied from interface: Admin

Query the status of the offline push by given kafka topic. TODO We use kafka topic to tracking the status now but in the further we should use jobId instead of kafka TODO topic. Right now each kafka topic only have one offline job. But in the further one kafka topic could be TODO assigned multiple jobs like data migration job etc.

Specified by:

getOffLinePushStatus in interface Admin

Returns:

the status of current offline push for the passed kafka topic

See Also:

• Admin.getOffLinePushStatus(String, String)

getOffLinePushStatus

public Admin.OfflinePushStatusInfo getOffLinePushStatus(String clusterName,
 String kafkaTopic,
 Optional<String> incrementalPushVersion,
 String region,
 String targetedRegions)

Specified by:

getOffLinePushStatus in interface Admin

getOverallPushStatus

protected static ExecutionStatus getOverallPushStatus(ExecutionStatus veniceStatus,
 ExecutionStatus daVinciStatus)

getKafkaBootstrapServers

public String getKafkaBootstrapServers(boolean isSSL)

Description copied from interface: Admin

Return the ssl or non-ssl bootstrap servers based on the given flag.

Specified by:

getKafkaBootstrapServers in interface Admin

Returns:

kafka bootstrap servers url, if there are multiple will be comma separated.

See Also:

• Admin.getKafkaBootstrapServers(boolean)

getRegionName

public String getRegionName()

Description copied from interface: Admin

Return the region name of this Admin

Specified by:

getRegionName in interface Admin

Returns:

the region name of this controller

getNativeReplicationKafkaBootstrapServerAddress

public String getNativeReplicationKafkaBootstrapServerAddress(String sourceFabric)

Specified by:

getNativeReplicationKafkaBootstrapServerAddress in interface Admin

Returns:

KafkaUrl for the given fabric.

See Also:

• ConfigKeys.CHILD_DATA_CENTER_KAFKA_URL_PREFIX

getNativeReplicationSourceFabric

public String getNativeReplicationSourceFabric(String clusterName,
 Store store,
 Optional<String> sourceGridFabric,
 Optional<String> emergencySourceRegion,
 String targetedRegions)

Source fabric selection priority: 1. Parent controller emergency source fabric config. 2. VPJ plugin targeted region config, however it will compute all selections based on the criteria below to select the source region. 3. VPJ plugin source grid fabric config. 4. Store level source fabric config. 5. Cluster level source fabric config.

Specified by:

getNativeReplicationSourceFabric in interface Admin

Returns:

the selected source fabric for a given store.

isSSLEnabledForPush

public boolean isSSLEnabledForPush(String clusterName,
 String storeName)

Description copied from interface: Admin

Return whether ssl is enabled for the given store for push.

Specified by:

isSSLEnabledForPush in interface Admin

See Also:

• Admin.isSSLEnabledForPush(String, String)

isSslToKafka

public boolean isSslToKafka()

Test if ssl is enabled to Kafka.

Specified by:

isSslToKafka in interface Admin

See Also:

• ConfigKeys.SSL_TO_KAFKA_LEGACY
• ConfigKeys.KAFKA_OVER_SSL

getTopicManager

public TopicManager getTopicManager()

Specified by:

getTopicManager in interface Admin

See Also:

• Admin.getTopicManager()

getTopicManager

public TopicManager getTopicManager(String pubSubServerAddress)

Specified by:

getTopicManager in interface Admin

See Also:

• Admin.getTopicManager(String)

isLeaderControllerFor

public boolean isLeaderControllerFor(String clusterName)

Description copied from interface: Admin

Check if this controller itself is the leader controller for a given cluster or not. Note that the controller can be either a parent controller or a child controller since a cluster must have a leader child controller and a leader parent controller. The point is not to be confused the concept of leader-standby with parent-child controller architecture.

Specified by:

isLeaderControllerFor in interface Admin

See Also:

• Admin.isLeaderControllerFor(String)

getAggregatedHealthStatus

public InstanceRemovableStatuses getAggregatedHealthStatus(String cluster,
 List<String> instances,
 List<String> toBeStoppedInstances,
 boolean isSSLEnabled)

Specified by:

getAggregatedHealthStatus in interface Admin

calculateNumberOfPartitions

public int calculateNumberOfPartitions(String clusterName,
 String storeName)

Calculate number of partition for given store.

Specified by:

calculateNumberOfPartitions in interface Admin

getReplicationFactor

public int getReplicationFactor(String clusterName,
 String storeName)

Specified by:

getReplicationFactor in interface Admin

Returns:

the replication factor of the given store.

getReplicas

public List<Replica> getReplicas(String clusterName,
 String kafkaTopic)

Specified by:

getReplicas in interface Admin

Returns:

a list of Replica created for the given resource.

getReplicasOfStorageNode

public List<Replica> getReplicasOfStorageNode(String cluster,
 String instanceId)

Specified by:

getReplicasOfStorageNode in interface Admin

See Also:

• Admin.getReplicasOfStorageNode(String, String)

isInstanceRemovable

public NodeRemovableResult isInstanceRemovable(String clusterName,
 String helixNodeId,
 List<String> lockedNodes)

Description copied from interface: Admin

Assuming all hosts identified by lockedNodes and their corresponding resources are unusable, is the given instance able to be removed out from the given cluster. For example, if there is only one online replica alive in this cluster which is hosted on the given instance. This instance should not be removed out of cluster, otherwise Venice will lose data. For detail criteria please refer to InstanceStatusDecider

Specified by:

isInstanceRemovable in interface Admin

Parameters:

clusterName - The cluster were the hosts belong.

helixNodeId - nodeId of helix participant. HOST_PORT.

lockedNodes - A list of helix nodeIds whose resources are assumed to be unusable (stopped).

See Also:

• Admin.isInstanceRemovable(String, String, List)

getLeaderController

public Instance getLeaderController(String clusterName)

Description copied from interface: Admin

Get instance of leader controller. If there is no leader controller for the given cluster, throw a VeniceException.

Specified by:

getLeaderController in interface Admin

See Also:

• Get the Venice controller leader for a storage cluster. We look at the external view of the controller cluster to find the Venice controller leader for a storage cluster. Because in both Helix as a library or Helix as a service (HaaS), the leader in the controller cluster external view is the Venice controller leader. During HaaS transition, controller leader property will become a HaaS controller, which is not the Venice controller that we want. Therefore, we don't refer to controller leader property to get leader controller.

addInstanceToAllowlist

public void addInstanceToAllowlist(String clusterName,
 String helixNodeId)

Add the given helix nodeId into the allowlist in ZK.

Specified by:

addInstanceToAllowlist in interface Admin

removeInstanceFromAllowList

public void removeInstanceFromAllowList(String clusterName,
 String helixNodeId)

Remove the given helix nodeId from the allowlist in ZK.

Specified by:

removeInstanceFromAllowList in interface Admin

getAllowlist

public Set<String> getAllowlist(String clusterName)

Specified by:

getAllowlist in interface Admin

Returns:

a list of all helix nodeIds in the allowlist for the given cluster from ZK.

killOfflinePush

public void killOfflinePush(String clusterName,
 String kafkaTopic,
 boolean isForcedKill)

Description copied from interface: Admin

Kill an offline push if it ran into errors or the corresponding version is being retired.

Specified by:

killOfflinePush in interface Admin

isForcedKill - should be set to true when killing the push job for retiring the corresponding version.

See Also:

• Admin.killOfflinePush(String, String, boolean)

deleteParticipantStoreKillMessage

public void deleteParticipantStoreKillMessage(String clusterName,
 String kafkaTopic)

Compose a ParticipantMessageKey message and execute a delete operation on the key to the cluster's participant store.

sendKillMessageToParticipantStore

public void sendKillMessageToParticipantStore(String clusterName,
 String kafkaTopic)

getStorageNodesStatus

public StorageNodeStatus getStorageNodesStatus(String clusterName,
 String instanceId)

Description copied from interface: Admin

Query and return the current status of the given storage node. The "storage node status" is composed by "status" of all replicas in that storage node. "status" is an integer value of Helix state:

• DROPPED=1
• ERROR=2
• OFFLINE=3
• BOOTSTRAP=4
• ONLINE=5

So this method will return a map, the key is the replica name which is composed by resource name and partitionId, and the value is the "status" of this replica.

Specified by:

getStorageNodesStatus in interface Admin

See Also:

• Admin.getStorageNodesStatus(String, boolean)

isStorageNodeNewerOrEqualTo

public boolean isStorageNodeNewerOrEqualTo(String clusterName,
 String instanceId,
 StorageNodeStatus oldStatus)

Description copied from interface: Admin

Compare the current storage node status and the given storage node status to check is the current one is "Newer" or "Equal" to the given one. Compare will go through each of replica in this storage node, if all their statuses values were larger or equal than the statuses value in the given storage node status, We say current storage node status is "Newer" or "Equal " to the given one.

Specified by:

isStorageNodeNewerOrEqualTo in interface Admin

See Also:

• Admin.isStorageNodeNewerOrEqualTo(String, String, StorageNodeStatus)

setAdminConsumerService

public void setAdminConsumerService(String clusterName,
 AdminConsumerService service)

Specified by:

setAdminConsumerService in interface Admin

See Also:

• Admin.setAdminConsumerService(String, AdminConsumerService)

skipAdminMessage

public void skipAdminMessage(String clusterName,
 long offset,
 boolean skipDIV)

Description copied from interface: Admin

The admin consumption task tries to deal with failures to process an admin message by retrying. If there is a message that cannot be processed for some reason, we will need to forcibly skip that message in order to unblock the task from consuming subsequent messages.

Specified by:

skipAdminMessage in interface Admin

skipDIV - tries to skip only the DIV check for the blocking message.

See Also:

• Admin.skipAdminMessage(String, long, boolean)

getLastSucceedExecutionId

public Long getLastSucceedExecutionId(String clusterName)

Description copied from interface: Admin

Get the id of the last succeed execution in this controller.

Specified by:

getLastSucceedExecutionId in interface Admin

See Also:

• Admin.getLastSucceedExecutionId(String)

getLastSucceededExecutionId

public Long getLastSucceededExecutionId(String clusterName,
 String storeName)

Get last succeeded execution id for a given store; if storeName is null, return the last succeeded execution id for the cluster

Parameters:

clusterName -

storeName -

Returns:

the last succeeded execution id or null if the cluster/store is invalid or the admin consumer service for the given cluster is not up and running yet.

getAdminCommandExecutionTracker

public Optional<AdminCommandExecutionTracker> getAdminCommandExecutionTracker(String clusterName)

Description copied from interface: Admin

Get the tracker used to track the execution of the admin command for the given cluster.

Specified by:

getAdminCommandExecutionTracker in interface Admin

See Also:

• Admin.getAdminCommandExecutionTracker(String)

getAdminTopicMetadata

public Map<String,Long> getAdminTopicMetadata(String clusterName,
 Optional<String> storeName)

Specified by:

getAdminTopicMetadata in interface Admin

Returns:

cluster-level execution id, offset and upstream offset. If store name is specified, it returns store-level execution id.

updateAdminTopicMetadata

public void updateAdminTopicMetadata(String clusterName,
 long executionId,
 Optional<String> storeName,
 Optional<Long> offset,
 Optional<Long> upstreamOffset)

Update cluster-level execution id, offset and upstream offset. If store name is specified, it updates the store-level execution id.

Specified by:

updateAdminTopicMetadata in interface Admin

getRoutersClusterConfig

public RoutersClusterConfig getRoutersClusterConfig(String clusterName)

Description copied from interface: Admin

Get the cluster level config for all routers.

Specified by:

getRoutersClusterConfig in interface Admin

See Also:

• Admin.getRoutersClusterConfig(String)

updateRoutersClusterConfig

public void updateRoutersClusterConfig(String clusterName,
 Optional<Boolean> isThrottlingEnable,
 Optional<Boolean> isQuotaRebalancedEnable,
 Optional<Boolean> isMaxCapacityProtectionEnabled,
 Optional<Integer> expectedRouterCount)

Description copied from interface: Admin

Update the cluster level for all routers.

Specified by:

updateRoutersClusterConfig in interface Admin

See Also:

• Admin.updateRoutersClusterConfig(String, Optional, Optional, Optional, Optional)

getAllStorePushStrategyForMigration

public Map<String,String> getAllStorePushStrategyForMigration()

Unsupported operation in the child controller.

Specified by:

getAllStorePushStrategyForMigration in interface Admin

setStorePushStrategyForMigration

public void setStorePushStrategyForMigration(String voldemortStoreName,
 String strategy)

Unsupported operation in the child controller.

Specified by:

setStorePushStrategyForMigration in interface Admin

discoverCluster

public Pair<String,String> discoverCluster(String storeName)

Description copied from interface: Admin

Find the cluster which the given store belongs to. Return the pair of the cluster name and the d2 service associated with that cluster.

Specified by:

discoverCluster in interface Admin

See Also:

• Admin.discoverCluster(String)

getServerD2Service

public String getServerD2Service(String clusterName)

Description copied from interface: Admin

Find the server d2 service associated with a given cluster name.

Specified by:

getServerD2Service in interface Admin

See Also:

• Admin.getServerD2Service(String)

findAllBootstrappingVersions

public Map<String,String> findAllBootstrappingVersions(String clusterName)

Description copied from interface: Admin

Find the store versions which have at least one bootstrap replica.

Specified by:

findAllBootstrappingVersions in interface Admin

See Also:

• TODO: With L/F we need to deprecate this function OR augment it to read the customized view as opposed to helix states

getVeniceWriterFactory

public VeniceWriterFactory getVeniceWriterFactory()

Specified by:

getVeniceWriterFactory in interface Admin

Returns:

a VeniceWriterFactory object used by the Venice controller to create the venice writer.

getPubSubConsumerAdapterFactory

public PubSubConsumerAdapterFactory getPubSubConsumerAdapterFactory()

Specified by:

getPubSubConsumerAdapterFactory in interface Admin

Returns:

a PubSubClientFactory object used by the Venice controller to create Pubsub clients.

getPubSubSSLProperties

public VeniceProperties getPubSubSSLProperties(String pubSubBrokerAddress)

Specified by:

getPubSubSSLProperties in interface Admin

stopMonitorOfflinePush

public void stopMonitorOfflinePush(String clusterName,
 String topic,
 boolean deletePushStatus,
 boolean isForcedDelete)

close

public void close()

Cause VeniceHelixAdmin and its associated services to stop executing.

Specified by:

close in interface Admin

Specified by:

close in interface AutoCloseable

Specified by:

close in interface Closeable

getHelixVeniceClusterResources

public HelixVeniceClusterResources getHelixVeniceClusterResources(String cluster)

Specified by:

getHelixVeniceClusterResources in interface Admin

Returns:

the aggregate resources required by controller to manage a Venice cluster.

getStoreConfigRepo

public HelixReadOnlyStoreConfigRepository getStoreConfigRepo()

Description copied from interface: Admin

Return a shared store config repository.

Specified by:

getStoreConfigRepo in interface Admin

isLeaderControllerOfControllerCluster

public boolean isLeaderControllerOfControllerCluster()

This function is used to detect whether current node is the leader controller of controller cluster. Be careful to use this function since it will talk to Zookeeper directly every time.

Specified by:

isLeaderControllerOfControllerCluster in interface Admin

Returns:

setStoreConfigForMigration

public void setStoreConfigForMigration(String storeName,
 String srcClusterName,
 String destClusterName)

Update "migrationDestCluster" and "migrationSrcCluster" fields of the "/storeConfigs/storeName" znode.

Parameters:

storeName - name of the store.

srcClusterName - name of the source cluster.

destClusterName - name of the destination cluster.

updateAclForStore

public void updateAclForStore(String clusterName,
 String storeName,
 String accessPermissions)

Description copied from interface: Admin

Provision a new set of ACL for a venice store and its associated kafka topic.

Specified by:

updateAclForStore in interface Admin

See Also:

• Admin.updateAclForStore(String, String, String)

getAclForStore

public String getAclForStore(String clusterName,
 String storeName)

Description copied from interface: Admin

Fetch the current set of ACL provisioned for a venice store and its associated kafka topic.

Specified by:

getAclForStore in interface Admin

Returns:

The string representation of the accessPermissions. It will return empty string in case store is not present.

See Also:

• Admin.getAclForStore(String, String)

deleteAclForStore

public void deleteAclForStore(String clusterName,
 String storeName)

Description copied from interface: Admin

Delete the current set of ACL provisioned for a venice store and its associated kafka topic.

Specified by:

deleteAclForStore in interface Admin

See Also:

• Admin.deleteAclForStore(String, String)

configureActiveActiveReplication

public void configureActiveActiveReplication(String clusterName,
 VeniceUserStoreType storeType,
 Optional<String> storeName,
 boolean enableActiveActiveReplicationForCluster,
 Optional<String> regionsFilter)

Description copied from interface: Admin

Enable/disable active active replications for certain stores (batch only, hybrid only, incremental push, hybrid or incremental push, all) in a cluster. If storeName is not empty, only the specified store might be updated.

Specified by:

configureActiveActiveReplication in interface Admin

See Also:

• Admin.configureActiveActiveReplication(String, VeniceUserStoreType, Optional, boolean, Optional)

getClusterStores

public ArrayList<StoreInfo> getClusterStores(String clusterName)

Description copied from interface: Admin

Return all stores in a cluster.

Specified by:

getClusterStores in interface Admin

Returns:

a list of StoreInfo of all stores in the specified cluster.

getClusterStaleStores

public Map<String,StoreDataAudit> getClusterStaleStores(String clusterName)

Specified by:

getClusterStaleStores in interface Admin

listStorePushInfo

public Map<String,RegionPushDetails> listStorePushInfo(String clusterName,
 String storeName,
 boolean isPartitionDetailEnabled)

Specified by:

listStorePushInfo in interface Admin

getRegionPushDetails

public RegionPushDetails getRegionPushDetails(String clusterName,
 String storeName,
 boolean isPartitionDetailAdded)

Specified by:

getRegionPushDetails in interface Admin

Returns:

RegionPushDetails object containing the specified store's push status.

retrievePushStatus

public OfflinePushStatus retrievePushStatus(String clusterName,
 StoreInfo store)

checkResourceCleanupBeforeStoreCreation

public void checkResourceCleanupBeforeStoreCreation(String clusterName,
 String storeName)

Description copied from interface: Admin

Check whether there are any resource left for the store creation in cluster: If there is any, this function should throw Exception.

Specified by:

checkResourceCleanupBeforeStoreCreation in interface Admin

See Also:

• Admin.checkResourceCleanupBeforeStoreCreation(String, String)

wipeCluster

public void wipeCluster(String clusterName,
 String fabric,
 Optional<String> storeName,
 Optional<Integer> versionNum)

Delete stores from the cluster including both store data and metadata.

The API provides the flexibility to delete a single store or a single version. Cluster name and fabric are required parameters, but store name and version number are optional. If store name is empty, all stores in the cluster are deleted.

Specified by:

wipeCluster in interface Admin

Parameters:

clusterName - name of the Venice cluster.

fabric - name of the fabric.

storeName - name of the to be deleted store, if value is absent, all stores in the cluster are deleted.

versionNum - the number of the version to be deleted, if present, only the specified version is deleted.

compareStore

public StoreComparisonInfo compareStore(String clusterName,
 String storeName,
 String fabricA,
 String fabricB)

Description copied from interface: Admin

Compare store metadata and version states between two fabrics.

Specified by:

compareStore in interface Admin

See Also:

• Admin.compareStore(String, String, String, String)

copyOverStoreSchemasAndConfigs

public StoreInfo copyOverStoreSchemasAndConfigs(String clusterName,
 String srcFabric,
 String destFabric,
 String storeName)

Specified by:

copyOverStoreSchemasAndConfigs in interface Admin

See Also:

• Admin.copyOverStoreSchemasAndConfigs(String, String, String, String)

isParent

public boolean isParent()

Description copied from interface: Admin

Check whether the controller works as a parent controller

Specified by:

isParent in interface Admin

Returns:

true if it works as a parent controller. Otherwise, return false.

See Also:

• Admin.isParent()

getParentControllerRegionState

public ParentControllerRegionState getParentControllerRegionState()

Description copied from interface: Admin

Return the state of the region of the parent controller.

Specified by:

getParentControllerRegionState in interface Admin

Returns:

ParentControllerRegionState.ACTIVE which means that the parent controller in the region is serving requests. Otherwise, return ParentControllerRegionState.PASSIVE

See Also:

• Admin.getParentControllerRegionState()

getChildDataCenterControllerUrlMap

public Map<String,String> getChildDataCenterControllerUrlMap(String clusterName)

Description copied from interface: Admin

Get child datacenter to child controller url mapping.

Specified by:

getChildDataCenterControllerUrlMap in interface Admin

Returns:

A map of child datacenter -> child controller url

See Also:

• Admin.getChildDataCenterControllerUrlMap(String)

getChildDataCenterControllerD2Map

public Map<String,String> getChildDataCenterControllerD2Map(String clusterName)

Description copied from interface: Admin

Get child datacenter to child controller d2 zk host mapping

Specified by:

getChildDataCenterControllerD2Map in interface Admin

Returns:

A map of child datacenter -> child controller d2 zk host

See Also:

• Admin.getChildDataCenterControllerD2Map(String)

getChildControllerD2ServiceName

public String getChildControllerD2ServiceName(String clusterName)

Description copied from interface: Admin

Get child datacenter controller d2 service name

Specified by:

getChildControllerD2ServiceName in interface Admin

Returns:

d2 service name

See Also:

• Admin.getChildControllerD2ServiceName(String)

getReadOnlyZKSharedSystemStoreRepository

public HelixReadOnlyZKSharedSystemStoreRepository getReadOnlyZKSharedSystemStoreRepository()

Description copied from interface: Admin

Return a shared read only store repository for zk shared stores.

Specified by:

getReadOnlyZKSharedSystemStoreRepository in interface Admin

See Also:

• Admin.getReadOnlyZKSharedSystemStoreRepository()

getReadOnlyZKSharedSchemaRepository

public HelixReadOnlyZKSharedSchemaRepository getReadOnlyZKSharedSchemaRepository()

Description copied from interface: Admin

Return a shared read only schema repository for zk shared stores.

Specified by:

getReadOnlyZKSharedSchemaRepository in interface Admin

See Also:

• Admin.getReadOnlyZKSharedSchemaRepository()

getMetaStoreWriter

public MetaStoreWriter getMetaStoreWriter()

Description copied from interface: Admin

Return a MetaStoreWriter, which can be shared across different Venice clusters.

Specified by:

getMetaStoreWriter in interface Admin

See Also:

• Admin.getMetaStoreWriter()

getMetaStoreReader

public MetaStoreReader getMetaStoreReader()

Specified by:

getMetaStoreReader in interface Admin

getEmergencySourceRegion

public Optional<String> getEmergencySourceRegion(@Nonnull
 String clusterName)

Description copied from interface: Admin

Return the emergency source region configuration.

Specified by:

getEmergencySourceRegion in interface Admin

See Also:

• Admin.getEmergencySourceRegion(String)

getAggregateRealTimeTopicSource

public Optional<String> getAggregateRealTimeTopicSource(String clusterName)

Description copied from interface: Admin

Return the source Kafka boostrap server url for aggregate real-time topic updates

Specified by:

getAggregateRealTimeTopicSource in interface Admin

See Also:

• Admin.getAggregateRealTimeTopicSource(String)

isActiveActiveReplicationEnabledInAllRegion

public boolean isActiveActiveReplicationEnabledInAllRegion(String clusterName,
 String storeName,
 boolean checkCurrentVersion)

Description copied from interface: Admin

Returns true if A/A replication is enabled in all child controller and parent controller. This is implemented only in parent controller. Otherwise, return false.

Specified by:

isActiveActiveReplicationEnabledInAllRegion in interface Admin

See Also:

• Admin.isActiveActiveReplicationEnabledInAllRegion(String, String, boolean)

getClustersLeaderOf

public List<String> getClustersLeaderOf()

Description copied from interface: Admin

Get a list of clusters this controller is a leader of.

Specified by:

getClustersLeaderOf in interface Admin

Returns:

a list of clusters this controller is a leader of.

See Also:

• Admin.getClustersLeaderOf()

getBackupVersionDefaultRetentionMs

public long getBackupVersionDefaultRetentionMs()

Description copied from interface: Admin

Returns default backup version retention time.

Specified by:

getBackupVersionDefaultRetentionMs in interface Admin

See Also:

• Admin.getBackupVersionDefaultRetentionMs()

getDefaultMaxRecordSizeBytes

public int getDefaultMaxRecordSizeBytes()

Specified by:

getDefaultMaxRecordSizeBytes in interface Admin

Returns:

The default value of VeniceWriter.maxRecordSizeBytes which is provided to the VPJ and Consumer as a controller config to dynamically control the setting per cluster.

See Also:

• Admin.getDefaultMaxRecordSizeBytes()

nodeReplicaReadiness

public Pair<NodeReplicasReadinessState,List<Replica>> nodeReplicaReadiness(String cluster,
 String helixNodeId)

Description copied from interface: Admin

helixNodeId nodeId of helix participant. HOST_PORT. Returns ture, if all current version replicas of the input node are ready to serve. false and all unready replicas otherwise.

Specified by:

nodeReplicaReadiness in interface Admin

See Also:

• Admin.nodeReplicaReadiness(String, String)

initiateDataRecovery

public void initiateDataRecovery(String clusterName,
 String storeName,
 int version,
 String sourceFabric,
 String destinationFabric,
 boolean copyAllVersionConfigs,
 Optional<Version> sourceFabricVersion)

Description copied from interface: Admin

Initiate data recovery for a store version given a source fabric.

Specified by:

initiateDataRecovery in interface Admin

Parameters:

clusterName - of the store.

storeName - of the store.

version - of the store.

sourceFabric - to be used as the source for data recovery.

copyAllVersionConfigs - a boolean to indicate whether all version configs should be copied from the source fabric or only the essential version configs and generate the rest based on destination fabric's Store configs.

sourceFabricVersion - source fabric's Version configs used to configure the recovering version in the destination fabric.

See Also:

• Admin.initiateDataRecovery(String, String, int, String, String, boolean, Optional)

prepareDataRecovery

public void prepareDataRecovery(String clusterName,
 String storeName,
 int version,
 String sourceFabric,
 String destinationFabric,
 Optional<Integer> sourceAmplificationFactor)

Description copied from interface: Admin

Prepare for data recovery in the destination fabric. The interested store version might have lingering states and resources in the destination fabric from previous failed attempts. Perform some basic checks to make sure the store version in the destination fabric is capable of performing data recovery and cleanup any lingering states and resources.

Specified by:

prepareDataRecovery in interface Admin

See Also:

• Admin.prepareDataRecovery(String, String, int, String, String, Optional)

isStoreVersionReadyForDataRecovery

public Pair<Boolean,String> isStoreVersionReadyForDataRecovery(String clusterName,
 String storeName,
 int version,
 String sourceFabric,
 String destinationFabric,
 Optional<Integer> sourceAmplificationFactor)

Description copied from interface: Admin

Check if the store version's previous states and resources are cleaned up and ready to start data recovery.

Specified by:

isStoreVersionReadyForDataRecovery in interface Admin

Returns:

whether is ready to start data recovery and the reason if it's not ready.

See Also:

• Admin.isStoreVersionReadyForDataRecovery(String, String, int, String, String, Optional)

isAdminTopicConsumptionEnabled

public boolean isAdminTopicConsumptionEnabled(String clusterName)

Description copied from interface: Admin

Return whether the admin consumption task is enabled for the passed cluster.

Specified by:

isAdminTopicConsumptionEnabled in interface Admin

See Also:

• Admin.isAdminTopicConsumptionEnabled(String)

getLargestUsedVersionFromStoreGraveyard

public int getLargestUsedVersionFromStoreGraveyard(String clusterName,
 String storeName)

Specified by:

getLargestUsedVersionFromStoreGraveyard in interface Admin

Returns:

the largest used version number for the given store from store graveyard.

createStoragePersona

public void createStoragePersona(String clusterName,
 String name,
 long quotaNumber,
 Set<String> storesToEnforce,
 Set<String> owners)

Specified by:

createStoragePersona in interface Admin

See Also:

• StoragePersonaRepository.addPersona(String, long, Set, Set)

getStoragePersona

public StoragePersona getStoragePersona(String clusterName,
 String name)

Specified by:

getStoragePersona in interface Admin

See Also:

• StoragePersonaRepository.getPersona(String)

deleteStoragePersona

public void deleteStoragePersona(String clusterName,
 String name)

Specified by:

deleteStoragePersona in interface Admin

See Also:

• StoragePersonaRepository.deletePersona(String)

updateStoragePersona

public void updateStoragePersona(String clusterName,
 String name,
 UpdateStoragePersonaQueryParams queryParams)

Specified by:

updateStoragePersona in interface Admin

See Also:

• StoragePersonaRepository.updatePersona(String, UpdateStoragePersonaQueryParams)

getPersonaAssociatedWithStore

public StoragePersona getPersonaAssociatedWithStore(String clusterName,
 String storeName)

Specified by:

getPersonaAssociatedWithStore in interface Admin

See Also:

• StoragePersonaRepository.getPersonaContainingStore(String)

getClusterStoragePersonas

public List<StoragePersona> getClusterStoragePersonas(String clusterName)

Specified by:

getClusterStoragePersonas in interface Admin

cleanupInstanceCustomizedStates

public List<String> cleanupInstanceCustomizedStates(String clusterName)

Description copied from interface: Admin

Scan through instance level customized states and remove any lingering ZNodes that are no longer relevant. This operation shouldn't be needed under normal circumstances. It's intended to cleanup ZNodes that failed to be deleted due to bugs and errors.

Specified by:

cleanupInstanceCustomizedStates in interface Admin

Parameters:

clusterName - to perform the cleanup.

Returns:

list of deleted ZNode paths.

getStoreGraveyard

public StoreGraveyard getStoreGraveyard()

Specified by:

getStoreGraveyard in interface Admin

removeStoreFromGraveyard

public void removeStoreFromGraveyard(String clusterName,
 String storeName)

Specified by:

removeStoreFromGraveyard in interface Admin

getPushStatusStoreReader

public PushStatusStoreReader getPushStatusStoreReader()

Specified by:

getPushStatusStoreReader in interface Admin

getPushStatusStoreWriter

public PushStatusStoreWriter getPushStatusStoreWriter()

Specified by:

getPushStatusStoreWriter in interface Admin

sendHeartbeatToSystemStore

public void sendHeartbeatToSystemStore(String clusterName,
 String storeName,
 long heartbeatTimeStamp)

Description copied from interface: Admin

Send a heartbeat timestamp to targeted system store.

Specified by:

sendHeartbeatToSystemStore in interface Admin

getHeartbeatFromSystemStore

public long getHeartbeatFromSystemStore(String clusterName,
 String systemStoreName)

Description copied from interface: Admin

Read the latest heartbeat timestamp from system store. If it failed to read from system store, this method should return -1.

Specified by:

getHeartbeatFromSystemStore in interface Admin

getSslFactory

public Optional<SSLFactory> getSslFactory()

isClusterWipeAllowed

public boolean isClusterWipeAllowed(String clusterName)

setPushJobDetailsStoreClient

public void setPushJobDetailsStoreClient(AvroSpecificStoreClient<PushJobStatusRecordKey,PushJobDetails> client)

getPubSubTopicRepository

public PubSubTopicRepository getPubSubTopicRepository()

Specified by:

getPubSubTopicRepository in interface Admin