blob: e3fae8a4185a69b376c16ea05be46011933cd55c [file] [log] [blame]
/*
* Licensed to the Apache Software Foundation (ASF) under one or more contributor license
* agreements. See the NOTICE file distributed with this work for additional information regarding
* copyright ownership. The ASF licenses this file to You under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance with the License. You may obtain a
* copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software distributed under the License
* is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express
* or implied. See the License for the specific language governing permissions and limitations under
* the License.
*/
package org.apache.geode.internal.cache.partitioned;
import java.io.DataInput;
import java.io.DataOutput;
import java.io.IOException;
import java.util.Set;
import org.apache.logging.log4j.Logger;
import org.apache.geode.CancelException;
import org.apache.geode.DataSerializer;
import org.apache.geode.cache.persistence.PartitionOfflineException;
import org.apache.geode.distributed.internal.ClusterDistributionManager;
import org.apache.geode.distributed.internal.DistributionManager;
import org.apache.geode.distributed.internal.DistributionMessage;
import org.apache.geode.distributed.internal.DistributionStats;
import org.apache.geode.distributed.internal.InternalDistributedSystem;
import org.apache.geode.distributed.internal.ReplyException;
import org.apache.geode.distributed.internal.ReplyMessage;
import org.apache.geode.distributed.internal.ReplyProcessor21;
import org.apache.geode.distributed.internal.membership.InternalDistributedMember;
import org.apache.geode.internal.Assert;
import org.apache.geode.internal.cache.ForceReattemptException;
import org.apache.geode.internal.cache.Node;
import org.apache.geode.internal.cache.PartitionedRegion;
import org.apache.geode.internal.cache.PartitionedRegionDataStore;
import org.apache.geode.internal.logging.LogService;
import org.apache.geode.internal.logging.log4j.LogMarker;
/**
* A request to manage a particular bucket
*
* @since GemFire 5.0
*
*/
public class ManageBucketMessage extends PartitionMessage {
private static final Logger logger = LogService.getLogger();
/** The key associated with the value that must be sent */
private int bucketId;
/** The value associated with the key that must be sent */
private int bucketSize;
/**
* Force the hosting of a bucket
*/
private boolean forceCreation;
/**
* Empty constructor to satisfy {@link DataSerializer} requirements
*/
public ManageBucketMessage() {}
private ManageBucketMessage(InternalDistributedMember recipient, int regionId,
ReplyProcessor21 processor, int bucketId, int bucketSize, boolean hostItNow) {
super(recipient, regionId, processor);
this.bucketId = bucketId;
this.bucketSize = bucketSize;
this.forceCreation = hostItNow;
}
public ManageBucketMessage(DataInput in) throws IOException, ClassNotFoundException {
fromData(in);
}
@Override
public int getProcessorType() {
return ClusterDistributionManager.WAITING_POOL_EXECUTOR;
}
/**
* Sends a PartitionedRegion manage bucket request to the recipient
*
* @param recipient the member to which the bucket manage request is sent
* @param r the PartitionedRegion to which the bucket belongs
* @param bucketId the unique identifier of the bucket
* @param bucketSize the size in bytes of the bucket
* @param forceCreation inform the recipient to accept the bucket appropriately ignoring maximums
* @return the processor used to fetch the returned Node if any
* @throws ForceReattemptException if the peer is no longer available
*/
public static NodeResponse send(InternalDistributedMember recipient, PartitionedRegion r,
int bucketId, int bucketSize, boolean forceCreation) throws ForceReattemptException {
Assert.assertTrue(recipient != null, "ManageBucketMessage NULL recipient");
NodeResponse p = new NodeResponse(r.getSystem(), recipient);
ManageBucketMessage m =
new ManageBucketMessage(recipient, r.getPRId(), p, bucketId, bucketSize, forceCreation);
m.setTransactionDistributed(r.getCache().getTxManager().isDistributed());
p.enableSevereAlertProcessing();
Set failures = r.getDistributionManager().putOutgoing(m);
if (failures != null && failures.size() > 0) {
throw new ForceReattemptException(
String.format("Failed sending < %s >", m));
}
return p;
}
@Override
public boolean canParticipateInTransaction() {
return false;
}
/**
* This method is called upon receipt and make the desired changes to the PartitionedRegion Note:
* It is very important that this message does NOT cause any deadlocks as the sender will wait
* indefinitely for the acknowledgement
*/
@Override
protected boolean operateOnPartitionedRegion(ClusterDistributionManager dm, PartitionedRegion r,
long startTime) {
if (logger.isTraceEnabled(LogMarker.DM_VERBOSE)) {
logger.trace(LogMarker.DM_VERBOSE, "ManageBucketMessage operateOnRegion: {}",
r.getFullPath());
}
// This is to ensure that initialization is complete before bucket creation request is
// serviced. BUGFIX for 35888
r.waitOnInitialization();
r.checkReadiness(); // Don't allow closed PartitionedRegions that have datastores to host
// buckets
PartitionedRegionDataStore prDs = r.getDataStore();
boolean managingBucket = prDs.handleManageBucketRequest(this.bucketId, this.bucketSize,
this.sender, this.forceCreation);
r.getPrStats().endPartitionMessagesProcessing(startTime);
if (managingBucket) {
// fix for bug 39356 - If the sender died while we were creating the bucket
// notify other nodes that they should invoke grabBackupBuckets to
// make copies of this bucket. Normally the sender would be responsible
// for creating those copies.
checkSenderStillAlive(r, getSender());
ManageBucketReplyMessage.sendAcceptance(getSender(), getProcessorId(), dm);
} else {
ManageBucketReplyMessage.sendRefusal(getSender(), getProcessorId(), dm);
}
return false;
}
/**
* Check that sender is still a participant in the partitioned region. If not, notify other nodes
* to create backup buckets
*/
private void checkSenderStillAlive(PartitionedRegion r, InternalDistributedMember sender) {
if (!r.getDistributionAdvisor().containsId(sender)) {
r.getRedundancyProvider().finishIncompleteBucketCreation(this.bucketId);
}
}
@Override
public int getDSFID() {
return PR_MANAGE_BUCKET_MESSAGE;
}
@Override
public void fromData(DataInput in) throws IOException, ClassNotFoundException {
super.fromData(in);
this.bucketId = in.readInt();
this.bucketSize = in.readInt();
this.forceCreation = in.readBoolean();
}
@Override
public void toData(DataOutput out) throws IOException {
super.toData(out);
out.writeInt(this.bucketId);
out.writeInt(this.bucketSize);
out.writeBoolean(this.forceCreation);
}
/**
* Assists the toString method in reporting the contents of this message
*
* @see PartitionMessage#toString()
*/
@Override
protected void appendFields(StringBuilder buff) {
super.appendFields(buff);
buff.append("; bucketId=").append(this.bucketId).append("; bucketSize=")
.append(this.bucketSize);
}
@Override
public boolean isSevereAlertCompatible() {
// since bucket management happens during entry operations, it
// must be severe-alert compatible
return true;
}
/**
* A class that contains the reply to a {@link ManageBucketMessage} message which contains the
* {@link Node} that has accepted to manage the bucket.
*
* @since GemFire 5.0
*/
public static class ManageBucketReplyMessage extends ReplyMessage {
protected boolean acceptedBucket;
/** true if the vm refused because it was still in initialization */
protected boolean notYetInitialized;
/**
* Empty constructor to conform to DataSerializable interface
*/
public ManageBucketReplyMessage() {}
public ManageBucketReplyMessage(DataInput in) throws IOException, ClassNotFoundException {
fromData(in);
}
private ManageBucketReplyMessage(int processorId, boolean accept, boolean initializing) {
this.processorId = processorId;
this.acceptedBucket = accept;
this.notYetInitialized = initializing;
}
/**
* Refuse the request to manage the bucket
*
* @param recipient the requesting node
* @param processorId the identity of the processor the requesting node is waiting on
* @param dm the distribution manager used to send the refusal
*/
public static void sendRefusal(InternalDistributedMember recipient, int processorId,
DistributionManager dm) {
Assert.assertTrue(recipient != null, "ManageBucketReplyMessage NULL reply message");
ManageBucketReplyMessage m = new ManageBucketReplyMessage(processorId, false, false);
m.setRecipient(recipient);
dm.putOutgoing(m);
}
/**
* Refuse the request to manage the bucket because the region is still being initialized
*
* @param recipient the requesting node
* @param processorId the identity of the processor the requesting node is waiting on
* @param dm the distribution manager used to send the acceptance message
*/
public static void sendStillInitializing(InternalDistributedMember recipient, int processorId,
DistributionManager dm) {
ManageBucketReplyMessage m = new ManageBucketReplyMessage(processorId, false, true);
m.setRecipient(recipient);
dm.putOutgoing(m);
}
/**
* Accept the request to manage the bucket
*
* @param recipient the requesting node
* @param processorId the identity of the processor the requesting node is waiting on
* @param dm the distribution manager used to send the acceptance message
*/
public static void sendAcceptance(InternalDistributedMember recipient, int processorId,
DistributionManager dm) {
Assert.assertTrue(recipient != null, "ManageBucketReplyMessage NULL reply message");
ManageBucketReplyMessage m = new ManageBucketReplyMessage(processorId, true, false);
m.setRecipient(recipient);
dm.putOutgoing(m);
}
/**
* Processes this message. This method is invoked by the receiver of the message.
*
* @param dm the distribution manager that is processing the message.
*/
@Override
public void process(final DistributionManager dm, final ReplyProcessor21 processor) {
final long startTime = getTimestamp();
if (logger.isTraceEnabled(LogMarker.DM_VERBOSE)) {
logger.trace(LogMarker.DM_VERBOSE,
"ManageBucketReplyMessage process invoking reply processor with processorId: {}",
this.processorId);
}
if (processor == null) {
if (logger.isTraceEnabled(LogMarker.DM_VERBOSE)) {
logger.trace(LogMarker.DM_VERBOSE, "ManageBucketReplyMessage processor not found");
}
return;
}
processor.process(this);
if (logger.isTraceEnabled(LogMarker.DM_VERBOSE)) {
logger.trace(LogMarker.DM_VERBOSE, "{} processed {}", processor, this);
}
dm.getStats().incReplyMessageTime(DistributionStats.getStatTime() - startTime);
}
@Override
public void toData(DataOutput out) throws IOException {
super.toData(out);
out.writeBoolean(this.acceptedBucket);
out.writeBoolean(this.notYetInitialized);
}
@Override
public int getDSFID() {
return PR_MANAGE_BUCKET_REPLY_MESSAGE;
}
@Override
public void fromData(DataInput in) throws IOException, ClassNotFoundException {
super.fromData(in);
this.acceptedBucket = in.readBoolean();
this.notYetInitialized = in.readBoolean();
}
@Override
public String toString() {
return new StringBuffer().append("ManageBucketReplyMessage ").append("processorid=")
.append(this.processorId).append(" accepted bucket=").append(this.acceptedBucket)
.append(" isInitializing=").append(this.notYetInitialized).toString();
}
}
/**
* A processor to capture the {@link Node} returned by {@link ManageBucketMessage}
*
* @since GemFire 5.0
*/
public static class NodeResponse extends ReplyProcessor21 {
/**
* the message that triggers return from waitForAcceptance. This will be null if the target
* member exited
*/
private volatile ManageBucketReplyMessage msg;
public NodeResponse(InternalDistributedSystem ds, InternalDistributedMember recipient) {
super(ds, recipient);
}
@Override
public void process(DistributionMessage msg) {
try {
if (msg instanceof ManageBucketReplyMessage) {
ManageBucketReplyMessage reply = (ManageBucketReplyMessage) msg;
this.msg = reply;
if (logger.isTraceEnabled(LogMarker.DM_VERBOSE)) {
logger.trace(LogMarker.DM_VERBOSE, "NodeResponse return value is {} isInitializing {}",
reply.acceptedBucket, reply.notYetInitialized);
}
} else {
Assert.assertTrue(msg instanceof ReplyMessage);
}
} finally {
super.process(msg);
}
}
/**
* Wait for the response to a {@link ManageBucketMessage} request.
*
* @return true if the node sent the request is managing the bucket
* @see org.apache.geode.internal.cache.PartitionedRegionDataStore#handleManageBucketRequest(int,
* int, InternalDistributedMember, boolean)
* @throws ForceReattemptException if the peer is no longer available
*/
public boolean waitForAcceptance() throws ForceReattemptException {
try {
waitForRepliesUninterruptibly();
} catch (ReplyException e) {
Throwable t = e.getCause();
if (t instanceof CancelException) {
logger.debug(
"NodeResponse got remote cancellation, throwing PartitionedRegionCommunication Exception. {}",
t.getMessage(), t);
throw new ForceReattemptException(
"NodeResponse got remote cancellation, throwing PartitionedRegionCommunication Exception.",
t);
}
if (t instanceof PRLocallyDestroyedException) {
logger.debug(
"NodeResponse got local destroy on the PartitionRegion , throwing ForceReattemptException. {}",
t.getMessage(), t);
throw new ForceReattemptException(
"NodeResponse got local destroy on the PartitionRegion , throwing ForceReattemptException.",
t);
}
if (t instanceof PartitionOfflineException) {
throw (PartitionOfflineException) t;
}
if (t instanceof ForceReattemptException) {
String msg =
"NodeResponse got ForceReattemptException due to local destroy on the PartitionRegion.";
logger.debug(msg, t);
throw (ForceReattemptException) t;
}
e.handleCause();
}
return (this.msg != null) && this.msg.acceptedBucket;
}
/**
* After a response has been returned from waitForAcceptance, this method may be used to see if
* the other vm rejected the bucket because it was still initializing.
*/
public boolean rejectedDueToInitialization() {
return (this.msg != null) && this.msg.notYetInitialized;
}
}
}