org.apache.cassandra.service

Class ActiveRepairService

java.lang.Object

org.apache.cassandra.service.ActiveRepairService

All Implemented Interfaces:

IEndpointStateChangeSubscriber, IFailureDetectionEventListener

public class ActiveRepairService
extends java.lang.Object
implements IEndpointStateChangeSubscriber, IFailureDetectionEventListener

ActiveRepairService is the starting point for manual "active" repairs. Each user triggered repair will correspond to one or multiple repair session, one for each token range to repair. On repair session might repair multiple column families. For each of those column families, the repair session will request merkle trees for each replica of the range being repaired, diff those trees upon receiving them, schedule the streaming ofthe parts to repair (based on the tree diffs) and wait for all those operation. See RepairSession for more details. The creation of a repair session is done through the submitRepairSession that returns a future on the completion of that session.

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	ActiveRepairService.ParentRepairSession We keep a ParentRepairSession around for the duration of the entire repair, for example, on a 256 token vnode rf=3 cluster we would have 768 RepairSession but only one ParentRepairSession.
static class	ActiveRepairService.Status Deprecated. this statuses are from the previous JMX notification service, which will be deprecated on 4.0. For statuses of the new notification service, see StreamEvent.ProgressEvent

Field Summary

Fields

Modifier and Type	Field and Description
static ActiveRepairService	instance
static CassandraVersion	SUPPORTS_GLOBAL_PREPARE_FLAG_VERSION
static long	UNREPAIRED_SSTABLE

Constructor Summary

Constructors

Constructor and Description
ActiveRepairService(IFailureDetector failureDetector, Gossiper gossiper)

Method Summary

Modifier and Type	Method and Description
void	beforeChange(java.net.InetAddress endpoint, EndpointState currentState, ApplicationState newStateKey, VersionedValue newValue)
void	convict(java.net.InetAddress ep, double phi) Something has happened to a remote node - if that node is a coordinator, we mark the parent repair session id as failed.
java.util.Set<SSTableReader>	currentlyRepairing(java.util.UUID cfId, java.util.UUID parentRepairSession)
com.google.common.util.concurrent.ListenableFuture<java.util.List<java.lang.Object>>	doAntiCompaction(java.util.UUID parentRepairSession, java.util.Collection<Range<Token>> successfulRanges) Submit anti-compaction jobs to CompactionManager.
com.google.common.util.concurrent.ListenableFuture	finishParentSession(java.util.UUID parentSession, java.util.Set<java.net.InetAddress> neighbors, java.util.Collection<Range<Token>> successfulRanges) Run final process of repair.
static java.util.Set<java.net.InetAddress>	getNeighbors(java.lang.String keyspaceName, java.util.Collection<Range<Token>> keyspaceLocalRanges, Range<Token> toRepair, java.util.Collection<java.lang.String> dataCenters, java.util.Collection<java.lang.String> hosts) Return all of the neighbors with whom we share the provided range.
ActiveRepairService.ParentRepairSession	getParentRepairSession(java.util.UUID parentSessionId)
void	handleMessage(java.net.InetAddress endpoint, RepairMessage message)
void	onAlive(java.net.InetAddress endpoint, EndpointState state)
void	onChange(java.net.InetAddress endpoint, ApplicationState state, VersionedValue value)
void	onDead(java.net.InetAddress endpoint, EndpointState state)
void	onJoin(java.net.InetAddress endpoint, EndpointState epState) Use to inform interested parties about the change in the state for specified endpoint
void	onRemove(java.net.InetAddress endpoint)
void	onRestart(java.net.InetAddress endpoint, EndpointState state) Called whenever a node is restarted.
java.util.UUID	prepareForRepair(java.util.UUID parentRepairSession, java.net.InetAddress coordinator, java.util.Set<java.net.InetAddress> endpoints, RepairOption options, java.util.List<ColumnFamilyStore> columnFamilyStores)
void	registerParentRepairSession(java.util.UUID parentRepairSession, java.net.InetAddress coordinator, java.util.List<ColumnFamilyStore> columnFamilyStores, java.util.Collection<Range<Token>> ranges, boolean isIncremental, long timestamp, boolean isGlobal)
ActiveRepairService.ParentRepairSession	removeParentRepairSession(java.util.UUID parentSessionId) called when the repair session is done - either failed or anticompaction has completed clears out any snapshots created by this repair
RepairSession	submitRepairSession(java.util.UUID parentRepairSession, java.util.Collection<Range<Token>> range, java.lang.String keyspace, RepairParallelism parallelismDegree, java.util.Set<java.net.InetAddress> endpoints, long repairedAt, com.google.common.util.concurrent.ListeningExecutorService executor, java.lang.String... cfnames) Requests repairs for the given keyspace and column families.
void	terminateSessions()

Methods inherited from class java.lang.Object

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

Field Detail

SUPPORTS_GLOBAL_PREPARE_FLAG_VERSION

public static CassandraVersion SUPPORTS_GLOBAL_PREPARE_FLAG_VERSION

instance

public static final ActiveRepairService instance

UNREPAIRED_SSTABLE

public static final long UNREPAIRED_SSTABLE

See Also:

Constant Field Values

Constructor Detail

ActiveRepairService

public ActiveRepairService(IFailureDetector failureDetector,
                           Gossiper gossiper)

Method Detail

submitRepairSession

public RepairSession submitRepairSession(java.util.UUID parentRepairSession,
                                         java.util.Collection<Range<Token>> range,
                                         java.lang.String keyspace,
                                         RepairParallelism parallelismDegree,
                                         java.util.Set<java.net.InetAddress> endpoints,
                                         long repairedAt,
                                         com.google.common.util.concurrent.ListeningExecutorService executor,
                                         java.lang.String... cfnames)

Requests repairs for the given keyspace and column families.

Returns:

Future for asynchronous call or null if there is no need to repair

terminateSessions

public void terminateSessions()

getNeighbors

public static java.util.Set<java.net.InetAddress> getNeighbors(java.lang.String keyspaceName,
                                                               java.util.Collection<Range<Token>> keyspaceLocalRanges,
                                                               Range<Token> toRepair,
                                                               java.util.Collection<java.lang.String> dataCenters,
                                                               java.util.Collection<java.lang.String> hosts)

Return all of the neighbors with whom we share the provided range.

Parameters:

keyspaceName - keyspace to repair

keyspaceLocalRanges - local-range for given keyspaceName

toRepair - token to repair

dataCenters - the data centers to involve in the repair

Returns:

neighbors with whom we share the provided range

prepareForRepair

public java.util.UUID prepareForRepair(java.util.UUID parentRepairSession,
                                       java.net.InetAddress coordinator,
                                       java.util.Set<java.net.InetAddress> endpoints,
                                       RepairOption options,
                                       java.util.List<ColumnFamilyStore> columnFamilyStores)

registerParentRepairSession

public void registerParentRepairSession(java.util.UUID parentRepairSession,
                                        java.net.InetAddress coordinator,
                                        java.util.List<ColumnFamilyStore> columnFamilyStores,
                                        java.util.Collection<Range<Token>> ranges,
                                        boolean isIncremental,
                                        long timestamp,
                                        boolean isGlobal)

currentlyRepairing

public java.util.Set<SSTableReader> currentlyRepairing(java.util.UUID cfId,
                                                       java.util.UUID parentRepairSession)

finishParentSession

public com.google.common.util.concurrent.ListenableFuture finishParentSession(java.util.UUID parentSession,
                                                                              java.util.Set<java.net.InetAddress> neighbors,
                                                                              java.util.Collection<Range<Token>> successfulRanges)

Run final process of repair. This removes all resources held by parent repair session, after performing anti compaction if necessary.

Parameters:

parentSession - Parent session ID

neighbors - Repair participants (not including self)

successfulRanges - Ranges that repaired successfully

getParentRepairSession

public ActiveRepairService.ParentRepairSession getParentRepairSession(java.util.UUID parentSessionId)

removeParentRepairSession

public ActiveRepairService.ParentRepairSession removeParentRepairSession(java.util.UUID parentSessionId)

called when the repair session is done - either failed or anticompaction has completed clears out any snapshots created by this repair

Parameters:

parentSessionId -

Returns:

doAntiCompaction

public com.google.common.util.concurrent.ListenableFuture<java.util.List<java.lang.Object>> doAntiCompaction(java.util.UUID parentRepairSession,
                                                                                                             java.util.Collection<Range<Token>> successfulRanges)

Submit anti-compaction jobs to CompactionManager. When all jobs are done, parent repair session is removed whether those are suceeded or not.

Parameters:

parentRepairSession - parent repair session ID

Returns:

Future result of all anti-compaction jobs.

handleMessage

public void handleMessage(java.net.InetAddress endpoint,
                          RepairMessage message)

onJoin

public void onJoin(java.net.InetAddress endpoint,
                   EndpointState epState)

Description copied from interface: IEndpointStateChangeSubscriber

Use to inform interested parties about the change in the state for specified endpoint

Specified by:

onJoin in interface IEndpointStateChangeSubscriber

Parameters:

endpoint - endpoint for which the state change occurred.

epState - state that actually changed for the above endpoint.

beforeChange

public void beforeChange(java.net.InetAddress endpoint,
                         EndpointState currentState,
                         ApplicationState newStateKey,
                         VersionedValue newValue)

Specified by:

beforeChange in interface IEndpointStateChangeSubscriber

onChange

public void onChange(java.net.InetAddress endpoint,
                     ApplicationState state,
                     VersionedValue value)

Specified by:

onChange in interface IEndpointStateChangeSubscriber

onAlive

public void onAlive(java.net.InetAddress endpoint,
                    EndpointState state)

Specified by:

onAlive in interface IEndpointStateChangeSubscriber

onDead

public void onDead(java.net.InetAddress endpoint,
                   EndpointState state)

Specified by:

onDead in interface IEndpointStateChangeSubscriber

onRemove

public void onRemove(java.net.InetAddress endpoint)

Specified by:

onRemove in interface IEndpointStateChangeSubscriber

onRestart

public void onRestart(java.net.InetAddress endpoint,
                      EndpointState state)

Description copied from interface: IEndpointStateChangeSubscriber

Called whenever a node is restarted. Note that there is no guarantee when that happens that the node was previously marked down. It will have only if state.isAlive() == false as state is from before the restarted node is marked up.

Specified by:

onRestart in interface IEndpointStateChangeSubscriber

convict

public void convict(java.net.InetAddress ep,
                    double phi)

Something has happened to a remote node - if that node is a coordinator, we mark the parent repair session id as failed. The fail marker is kept in the map for 24h to make sure that if the coordinator does not agree that the repair failed, we need to fail the entire repair session

Specified by:

convict in interface IFailureDetectionEventListener

Parameters:

ep - endpoint to be convicted

phi - the value of phi with with ep was convicted