org.apache.cassandra.repair

Class RepairSession

java.lang.Object

org.apache.cassandra.utils.concurrent.AbstractFuture<V>

org.apache.cassandra.utils.concurrent.AsyncFuture<RepairSessionResult>

org.apache.cassandra.repair.RepairSession

All Implemented Interfaces:

com.google.common.util.concurrent.ListenableFuture<RepairSessionResult>, java.util.concurrent.Future<RepairSessionResult>, IEndpointStateChangeSubscriber, IFailureDetectionEventListener, LocalSessions.Listener, Awaitable

public class RepairSession
extends AsyncFuture<RepairSessionResult>
implements IEndpointStateChangeSubscriber, IFailureDetectionEventListener, LocalSessions.Listener

Coordinates the (active) repair of a list of non overlapping token ranges. A given RepairSession repairs a set of replicas for a given set of ranges on a list of column families. For each of the column family to repair, RepairSession creates a RepairJob that handles the repair of that CF. A given RepairJob has the 3 main phases:

1. Paxos repair: unfinished paxos operations in the range/keyspace/table are first completed
2. Validation phase: the job requests merkle trees from each of the replica involves (ValidationTask) and waits until all trees are received (in validationComplete()).
3. Synchronization phase: once all trees are received, the job compares each tree with all the others. If there is difference between 2 trees, the differences between the 2 endpoints will be streamed with a SyncTask.

The job is done once all its SyncTasks are done (i.e. have either computed no differences or the streaming they started is done (syncComplete())). A given session will execute the first phase (validation phase) of each of it's job sequentially. In other words, it will start the first job and only start the next one once that first job validation phase is complete. This is done so that the replica only create one merkle tree per range at a time, which is our way to ensure that such creation starts roughly at the same time on every node (see CASSANDRA-2816). However the synchronization phases are allowed to run concurrently (with each other and with validation phases). A given RepairJob has 2 modes: either sequential or not (RepairParallelism). If sequential, it will requests merkle tree creation from each replica in sequence (though in that case we still first send a message to each node to flush and snapshot data so each merkle tree creation is still done on similar data, even if the actual creation is not done simulatneously). If not sequential, all merkle tree are requested in parallel. Similarly, if a job is sequential, it will handle one SymmetricSyncTask at a time, but will handle all of them in parallel otherwise.

Nested Class Summary

Nested classes/interfaces inherited from interface org.apache.cassandra.utils.concurrent.Awaitable

Awaitable.AbstractAwaitable, Awaitable.AsyncAwaitable, Awaitable.Defaults, Awaitable.SyncAwaitable

Field Summary

Fields

Modifier and Type	Field and Description
boolean	isIncremental Range to repair
boolean	optimiseStreams
RepairParallelism	parallelismDegree
boolean	paxosOnly
PreviewKind	previewKind
boolean	pullRepair
boolean	repairPaxos
SessionState	state
ExecutorPlus	taskExecutor
Scheduler	validationScheduler

Fields inherited from class org.apache.cassandra.utils.concurrent.AbstractFuture

CANCELLED, UNCANCELLABLE, UNSET

Constructor Summary

Constructors

Constructor and Description
RepairSession(TimeUUID parentRepairSession, Scheduler validationScheduler, CommonRange commonRange, java.lang.String keyspace, RepairParallelism parallelismDegree, boolean isIncremental, boolean pullRepair, PreviewKind previewKind, boolean optimiseStreams, boolean repairPaxos, boolean paxosOnly, java.lang.String... cfnames) Create new repair session.

Method Summary

Modifier and Type	Method and Description
void	convict(InetAddressAndPort endpoint, double phi) Convict the specified endpoint.
protected ExecutorPlus	createExecutor()
java.util.Collection<InetAddressAndPort>	endpoints()
void	forceShutdown(java.lang.Throwable reason) clear all RepairJobs and terminate this session.
TimeUUID	getId()
void	onIRStateChange(LocalSession session)
void	onRemove(InetAddressAndPort endpoint)
void	onRestart(InetAddressAndPort endpoint, EndpointState epState) Called whenever a node is restarted.
java.util.Collection<Range<Token>>	ranges()
void	start(ExecutorPlus executor) Start RepairJob on given ColumnFamilies.
void	syncComplete(RepairJobDesc desc, SyncNodePair nodes, boolean success, java.util.List<SessionSummary> summaries) Notify this session that sync completed/failed with given SyncNodePair.
void	terminate()
void	trackSyncCompletion(Pair<RepairJobDesc,SyncNodePair> key, CompletableRemoteSyncTask task)
void	trackValidationCompletion(Pair<RepairJobDesc,InetAddressAndPort> key, ValidationTask task)
void	validationComplete(RepairJobDesc desc, InetAddressAndPort endpoint, MerkleTrees trees) Receive merkle tree response or failed response from endpoint for current repair job.

Methods inherited from class org.apache.cassandra.utils.concurrent.AsyncFuture

await, awaitUntil, flatMap, map

Methods inherited from class org.apache.cassandra.utils.concurrent.AbstractFuture

addCallback, addCallback, addCallback, addCallback, addCallback, addCallback, addListener, addListener, addListener, addListeners, await, awaitThrowUncheckedOnInterrupt, awaitThrowUncheckedOnInterrupt, awaitUninterruptibly, awaitUninterruptibly, awaitUntilThrowUncheckedOnInterrupt, awaitUntilUninterruptibly, cancel, cause, description, flatMap, get, get, getNow, getWhenDone, isCancellable, isCancelled, isDone, isSuccess, isUncancellable, map, map, notifyExecutor, removeListener, removeListeners, setUncancellable, setUncancellableExclusive, toString, tryFailure, trySuccess

Methods inherited from class java.lang.Object

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

Methods inherited from interface org.apache.cassandra.gms.IEndpointStateChangeSubscriber

beforeChange, onAlive, onChange, onDead, onJoin

Methods inherited from interface org.apache.cassandra.utils.concurrent.Future

await, awaitUninterruptibly, flatMap, rethrowIfFailed, sync, syncThrowUncheckedOnInterrupt, syncUninterruptibly

Field Detail

state

public final SessionState state

parallelismDegree

public final RepairParallelism parallelismDegree

pullRepair

public final boolean pullRepair

isIncremental

public final boolean isIncremental

Range to repair

previewKind

public final PreviewKind previewKind

repairPaxos

public final boolean repairPaxos

paxosOnly

public final boolean paxosOnly

taskExecutor

public final ExecutorPlus taskExecutor

optimiseStreams

public final boolean optimiseStreams

validationScheduler

public final Scheduler validationScheduler

Constructor Detail

RepairSession

public RepairSession(TimeUUID parentRepairSession,
                     Scheduler validationScheduler,
                     CommonRange commonRange,
                     java.lang.String keyspace,
                     RepairParallelism parallelismDegree,
                     boolean isIncremental,
                     boolean pullRepair,
                     PreviewKind previewKind,
                     boolean optimiseStreams,
                     boolean repairPaxos,
                     boolean paxosOnly,
                     java.lang.String... cfnames)

Create new repair session.

Parameters:

parentRepairSession - the parent sessions id

commonRange - ranges to repair

keyspace - name of keyspace

parallelismDegree - specifies the degree of parallelism when calculating the merkle trees

pullRepair - true if the repair should be one way (from remote host to this host and only applicable between two hosts--see RepairOption)

repairPaxos - true if incomplete paxos operations should be completed as part of repair

paxosOnly - true if we should only complete paxos operations, not run a normal repair

cfnames - names of columnfamilies

Method Detail

createExecutor

protected ExecutorPlus createExecutor()

getId

public TimeUUID getId()

ranges

public java.util.Collection<Range<Token>> ranges()

endpoints

public java.util.Collection<InetAddressAndPort> endpoints()

trackValidationCompletion

public void trackValidationCompletion(Pair<RepairJobDesc,InetAddressAndPort> key,
                                      ValidationTask task)

trackSyncCompletion

public void trackSyncCompletion(Pair<RepairJobDesc,SyncNodePair> key,
                                CompletableRemoteSyncTask task)

validationComplete

public void validationComplete(RepairJobDesc desc,
                               InetAddressAndPort endpoint,
                               MerkleTrees trees)

Receive merkle tree response or failed response from endpoint for current repair job.

Parameters:

desc - repair job description

endpoint - endpoint that sent merkle tree

trees - calculated merkle trees, or null if validation failed

syncComplete

public void syncComplete(RepairJobDesc desc,
                         SyncNodePair nodes,
                         boolean success,
                         java.util.List<SessionSummary> summaries)

Notify this session that sync completed/failed with given SyncNodePair.

Parameters:

desc - synced repair job

nodes - nodes that completed sync

success - true if sync succeeded

start

public void start(ExecutorPlus executor)

Start RepairJob on given ColumnFamilies. This first validates if all replica are available, and if they are, creates RepairJobs and submit to run on given executor.

Parameters:

executor - Executor to run validation

terminate

public void terminate()

forceShutdown

public void forceShutdown(java.lang.Throwable reason)

clear all RepairJobs and terminate this session.

Parameters:

reason - Cause of error for shutdown

onRemove

public void onRemove(InetAddressAndPort endpoint)

Specified by:

onRemove in interface IEndpointStateChangeSubscriber

onRestart

public void onRestart(InetAddressAndPort endpoint,
                      EndpointState epState)

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(InetAddressAndPort endpoint,
                    double phi)

Description copied from interface: IFailureDetectionEventListener

Convict the specified endpoint.

Specified by:

convict in interface IFailureDetectionEventListener

Parameters:

endpoint - endpoint to be convicted

phi - the value of phi with with ep was convicted

onIRStateChange

public void onIRStateChange(LocalSession session)

Specified by:

onIRStateChange in interface LocalSessions.Listener