org.elasticsearch.indices.recovery

Class RecoverySourceHandler

java.lang.Object

org.elasticsearch.indices.recovery.RecoverySourceHandler

Direct Known Subclasses:

SharedFSRecoverySourceHandler

public class RecoverySourceHandler
extends java.lang.Object

RecoverySourceHandler handles the three phases of shard recovery, which is everything relating to copying the segment files as well as sending translog operations across the wire once the segments have been copied. Note: There is always one source handler per recovery that handles all the file and translog transfer. This handler is completely isolated from other recoveries while the RateLimiter passed via RecoverySettings is shared across recoveries originating from this nodes to throttle the number bytes send during file transfer. The transaction log phase bypasses the rate limiter entirely.

Field Summary

Fields

Modifier and Type	Field and Description
protected org.apache.logging.log4j.Logger	logger
protected org.elasticsearch.indices.recovery.RecoveryResponse	response

Constructor Summary

Constructors

Constructor and Description
RecoverySourceHandler(IndexShard shard, RecoveryTargetHandler recoveryTarget, StartRecoveryRequest request, java.util.function.Supplier<java.lang.Long> currentClusterStateVersionSupplier, java.util.function.Function<java.lang.String,Releasable> delayNewRecoveries, int fileChunkSizeInBytes, org.apache.logging.log4j.Logger logger)

Method Summary

Modifier and Type	Method and Description
void	cancel(java.lang.String reason) Cancels the recovery and interrupts all eligible threads.
protected void	failEngine(java.io.IOException cause)
void	finalizeRecovery() finalizes the recovery process
void	phase1(org.apache.lucene.index.IndexCommit snapshot, Translog.View translogView) Perform phase1 of the recovery operations.
void	phase2(Translog.Snapshot snapshot) Perform phase2 of the recovery process
protected void	prepareTargetForTranslog(int totalTranslogOps, long maxUnsafeAutoIdTimestamp)
org.elasticsearch.indices.recovery.RecoveryResponse	recoverToTarget() performs the recovery from the local engine to the target
protected int	sendSnapshot(Translog.Snapshot snapshot) Send the given snapshot's operations to this handler's target node.
java.lang.String	toString()

Methods inherited from class java.lang.Object

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

Field Detail

logger

protected final org.apache.logging.log4j.Logger logger

response

protected final org.elasticsearch.indices.recovery.RecoveryResponse response

Constructor Detail

RecoverySourceHandler

public RecoverySourceHandler(IndexShard shard,
                             RecoveryTargetHandler recoveryTarget,
                             StartRecoveryRequest request,
                             java.util.function.Supplier<java.lang.Long> currentClusterStateVersionSupplier,
                             java.util.function.Function<java.lang.String,Releasable> delayNewRecoveries,
                             int fileChunkSizeInBytes,
                             org.apache.logging.log4j.Logger logger)

Method Detail

recoverToTarget

public org.elasticsearch.indices.recovery.RecoveryResponse recoverToTarget()
                                                                    throws java.io.IOException

performs the recovery from the local engine to the target

Throws:

java.io.IOException

phase1

public void phase1(org.apache.lucene.index.IndexCommit snapshot,
                   Translog.View translogView)

Perform phase1 of the recovery operations. Once this IndexCommit snapshot has been performed no commit operations (files being fsync'd) are effectively allowed on this index until all recovery phases are done

Phase1 examines the segment files on the target node and copies over the segments that are missing. Only segments that have the same size and checksum can be reused

prepareTargetForTranslog

protected void prepareTargetForTranslog(int totalTranslogOps,
                                        long maxUnsafeAutoIdTimestamp)
                                 throws java.io.IOException

Throws:

java.io.IOException

phase2

public void phase2(Translog.Snapshot snapshot)

Perform phase2 of the recovery process

Phase2 takes a snapshot of the current translog *without* acquiring the write lock (however, the translog snapshot is a point-in-time view of the translog). It then sends each translog operation to the target node so it can be replayed into the new shard.

finalizeRecovery

public void finalizeRecovery()

finalizes the recovery process

sendSnapshot

protected int sendSnapshot(Translog.Snapshot snapshot)

Send the given snapshot's operations to this handler's target node.

Operations are bulked into a single request depending on an operation count limit or size-in-bytes limit

Returns:

the total number of translog operations that were sent

cancel

public void cancel(java.lang.String reason)

Cancels the recovery and interrupts all eligible threads.

toString

public java.lang.String toString()

Overrides:

toString in class java.lang.Object

failEngine

protected void failEngine(java.io.IOException cause)