/*
* Licensed to Elasticsearch under one or more contributor
* license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright
* ownership. Elasticsearch 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.elasticsearch.cluster.routing.allocation;
import com.carrotsearch.hppc.cursors.ObjectCursor;
import com.google.common.collect.ImmutableList;
import com.google.common.collect.Lists;
import org.elasticsearch.ElasticsearchException;
import org.elasticsearch.ElasticsearchIllegalStateException;
import org.elasticsearch.cluster.ClusterInfoService;
import org.elasticsearch.cluster.ClusterState;
import org.elasticsearch.cluster.node.DiscoveryNode;
import org.elasticsearch.cluster.routing.*;
import org.elasticsearch.cluster.routing.allocation.allocator.ShardsAllocators;
import org.elasticsearch.cluster.routing.allocation.command.AllocationCommands;
import org.elasticsearch.cluster.routing.allocation.decider.AllocationDeciders;
import org.elasticsearch.cluster.routing.allocation.decider.Decision;
import org.elasticsearch.common.component.AbstractComponent;
import org.elasticsearch.common.inject.Inject;
import org.elasticsearch.common.settings.Settings;
import java.util.ArrayList;
import java.util.Iterator;
import java.util.List;
import static org.elasticsearch.cluster.routing.ShardRoutingState.INITIALIZING;
import static org.elasticsearch.cluster.routing.ShardRoutingState.RELOCATING;
/**
* This service manages the node allocation of a cluster. For this reason the
* {@link AllocationService} keeps {@link AllocationDeciders} to choose nodes
* for shard allocation. This class also manages new nodes joining the cluster
* and rerouting of shards.
*/
public class AllocationService extends AbstractComponent {
private final AllocationDeciders allocationDeciders;
private final ClusterInfoService clusterInfoService;
private final ShardsAllocators shardsAllocators;
@Inject
public AllocationService(Settings settings, AllocationDeciders allocationDeciders, ShardsAllocators shardsAllocators, ClusterInfoService clusterInfoService) {
super(settings);
this.allocationDeciders = allocationDeciders;
this.shardsAllocators = shardsAllocators;
this.clusterInfoService = clusterInfoService;
}
/**
* Applies the started shards. Note, shards can be called several times within this method.
* <p/>
* <p>If the same instance of the routing table is returned, then no change has been made.</p>
*/
public RoutingAllocation.Result applyStartedShards(ClusterState clusterState, List<? extends ShardRouting> startedShards) {
return applyStartedShards(clusterState, startedShards, true);
}
public RoutingAllocation.Result applyStartedShards(ClusterState clusterState, List<? extends ShardRouting> startedShards, boolean withReroute) {
RoutingNodes routingNodes = clusterState.routingNodes();
// shuffle the unassigned nodes, just so we won't have things like poison failed shards
routingNodes.unassigned().shuffle();
StartedRerouteAllocation allocation = new StartedRerouteAllocation(allocationDeciders, routingNodes, clusterState.nodes(), startedShards, clusterInfoService.getClusterInfo());
boolean changed = applyStartedShards(routingNodes, startedShards);
if (!changed) {
return new RoutingAllocation.Result(false, clusterState.routingTable());
}
shardsAllocators.applyStartedShards(allocation);
if (withReroute) {
reroute(allocation);
}
return new RoutingAllocation.Result(true, new RoutingTable.Builder().updateNodes(routingNodes).build().validateRaiseException(clusterState.metaData()));
}
public RoutingAllocation.Result applyFailedShard(ClusterState clusterState, ShardRouting failedShard) {
return applyFailedShards(clusterState, ImmutableList.of(failedShard));
}
/**
* Applies the failed shards. Note, shards can be called several times within this method.
* <p/>
* <p>If the same instance of the routing table is returned, then no change has been made.</p>
*/
public RoutingAllocation.Result applyFailedShards(ClusterState clusterState, List<ShardRouting> failedShards) {
RoutingNodes routingNodes = clusterState.routingNodes();
// shuffle the unassigned nodes, just so we won't have things like poison failed shards
routingNodes.unassigned().shuffle();
FailedRerouteAllocation allocation = new FailedRerouteAllocation(allocationDeciders, routingNodes, clusterState.nodes(), failedShards, clusterInfoService.getClusterInfo());
boolean changed = false;
for (ShardRouting failedShard : failedShards) {
changed |= applyFailedShard(allocation, failedShard, true);
}
if (!changed) {
return new RoutingAllocation.Result(false, clusterState.routingTable());
}
shardsAllocators.applyFailedShards(allocation);
reroute(allocation);
return new RoutingAllocation.Result(true, new RoutingTable.Builder().updateNodes(routingNodes).build().validateRaiseException(clusterState.metaData()));
}
public RoutingAllocation.Result reroute(ClusterState clusterState, AllocationCommands commands) {
return reroute(clusterState, commands, false);
}
public RoutingAllocation.Result reroute(ClusterState clusterState, AllocationCommands commands, boolean explain) throws ElasticsearchException {
RoutingNodes routingNodes = clusterState.routingNodes();
// we don't shuffle the unassigned shards here, to try and get as close as possible to
// a consistent result of the effect the commands have on the routing
// this allows systems to dry run the commands, see the resulting cluster state, and act on it
RoutingAllocation allocation = new RoutingAllocation(allocationDeciders, routingNodes, clusterState.nodes(), clusterInfoService.getClusterInfo());
// don't short circuit deciders, we want a full explanation
allocation.debugDecision(true);
// we ignore disable allocation, because commands are explicit
allocation.ignoreDisable(true);
RoutingExplanations explanations = commands.execute(allocation, explain);
// we revert the ignore disable flag, since when rerouting, we want the original setting to take place
allocation.ignoreDisable(false);
// the assumption is that commands will move / act on shards (or fail through exceptions)
// so, there will always be shard "movements", so no need to check on reroute
reroute(allocation);
return new RoutingAllocation.Result(true, new RoutingTable.Builder().updateNodes(routingNodes).build().validateRaiseException(clusterState.metaData()), explanations);
}
/**
* Reroutes the routing table based on the live nodes.
* <p/>
* <p>If the same instance of the routing table is returned, then no change has been made.
*/
public RoutingAllocation.Result reroute(ClusterState clusterState) {
return reroute(clusterState, false);
}
/**
* Reroutes the routing table based on the live nodes.
* <p/>
* <p>If the same instance of the routing table is returned, then no change has been made.
*/
public RoutingAllocation.Result reroute(ClusterState clusterState, boolean debug) {
RoutingNodes routingNodes = clusterState.routingNodes();
// shuffle the unassigned nodes, just so we won't have things like poison failed shards
routingNodes.unassigned().shuffle();
RoutingAllocation allocation = new RoutingAllocation(allocationDeciders, routingNodes, clusterState.nodes(), clusterInfoService.getClusterInfo());
allocation.debugDecision(debug);
if (!reroute(allocation)) {
return new RoutingAllocation.Result(false, clusterState.routingTable());
}
return new RoutingAllocation.Result(true, new RoutingTable.Builder().updateNodes(routingNodes).build().validateRaiseException(clusterState.metaData()));
}
/**
* Only handles reroute but *without* any reassignment of unassigned shards or rebalancing. Does
* make sure to handle removed nodes, but only moved the shards to UNASSIGNED, does not reassign
* them.
*/
public RoutingAllocation.Result rerouteWithNoReassign(ClusterState clusterState) {
return rerouteWithNoReassign(clusterState, false);
}
/**
* Only handles reroute but *without* any reassignment of unassigned shards or rebalancing. Does
* make sure to handle removed nodes, but only moved the shards to UNASSIGNED, does not reassign
* them.
*/
public RoutingAllocation.Result rerouteWithNoReassign(ClusterState clusterState, boolean debug) {
RoutingNodes routingNodes = clusterState.routingNodes();
// shuffle the unassigned nodes, just so we won't have things like poison failed shards
routingNodes.unassigned().shuffle();
RoutingAllocation allocation = new RoutingAllocation(allocationDeciders, routingNodes, clusterState.nodes(), clusterInfoService.getClusterInfo());
allocation.debugDecision(debug);
boolean changed = false;
// first, clear from the shards any node id they used to belong to that is now dead
changed |= deassociateDeadNodes(allocation);
// create a sorted list of from nodes with least number of shards to the maximum ones
applyNewNodes(allocation);
// elect primaries *before* allocating unassigned, so backups of primaries that failed
// will be moved to primary state and not wait for primaries to be allocated and recovered (*from gateway*)
changed |= electPrimariesAndUnassignedDanglingReplicas(allocation);
if (!changed) {
return new RoutingAllocation.Result(false, clusterState.routingTable());
}
return new RoutingAllocation.Result(true, new RoutingTable.Builder().updateNodes(routingNodes).build().validateRaiseException(clusterState.metaData()));
}
private boolean reroute(RoutingAllocation allocation) {
boolean changed = false;
// first, clear from the shards any node id they used to belong to that is now dead
changed |= deassociateDeadNodes(allocation);
// create a sorted list of from nodes with least number of shards to the maximum ones
applyNewNodes(allocation);
// elect primaries *before* allocating unassigned, so backups of primaries that failed
// will be moved to primary state and not wait for primaries to be allocated and recovered (*from gateway*)
changed |= electPrimariesAndUnassignedDanglingReplicas(allocation);
// now allocate all the unassigned to available nodes
if (allocation.routingNodes().hasUnassigned()) {
changed |= shardsAllocators.allocateUnassigned(allocation);
// elect primaries again, in case this is needed with unassigned allocation
changed |= electPrimariesAndUnassignedDanglingReplicas(allocation);
}
// move shards that no longer can be allocated
changed |= moveShards(allocation);
// rebalance
changed |= shardsAllocators.rebalance(allocation);
assert RoutingNodes.assertShardStats(allocation.routingNodes());
return changed;
}
private boolean moveShards(RoutingAllocation allocation) {
boolean changed = false;
// create a copy of the shards interleaving between nodes, and check if they can remain
List<MutableShardRouting> shards = new ArrayList<>();
int index = 0;
boolean found = true;
final RoutingNodes routingNodes = allocation.routingNodes();
while (found) {
found = false;
for (RoutingNode routingNode : routingNodes) {
if (index >= routingNode.size()) {
continue;
}
found = true;
shards.add(routingNode.get(index));
}
index++;
}
for (int i = 0; i < shards.size(); i++) {
MutableShardRouting shardRouting = shards.get(i);
// we can only move started shards...
if (!shardRouting.started()) {
continue;
}
final RoutingNode routingNode = routingNodes.node(shardRouting.currentNodeId());
Decision decision = allocation.deciders().canRemain(shardRouting, routingNode, allocation);
if (decision.type() == Decision.Type.NO) {
logger.debug("[{}][{}] allocated on [{}], but can no longer be allocated on it, moving...", shardRouting.index(), shardRouting.id(), routingNode.node());
boolean moved = shardsAllocators.move(shardRouting, routingNode, allocation);
if (!moved) {
logger.debug("[{}][{}] can't move", shardRouting.index(), shardRouting.id());
} else {
changed = true;
}
}
}
return changed;
}
private boolean electPrimariesAndUnassignedDanglingReplicas(RoutingAllocation allocation) {
boolean changed = false;
RoutingNodes routingNodes = allocation.routingNodes();
if (!routingNodes.hasUnassignedPrimaries()) {
// move out if we don't have unassigned primaries
return changed;
}
// go over and remove dangling replicas that are initializing for primary shards
List<ShardRouting> shardsToFail = Lists.newArrayList();
for (MutableShardRouting shardEntry : routingNodes.unassigned()) {
if (shardEntry.primary()) {
for (MutableShardRouting routing : routingNodes.assignedShards(shardEntry)) {
if (!routing.primary() && routing.initializing()) {
shardsToFail.add(routing);
}
}
}
}
for (ShardRouting shardToFail : shardsToFail) {
changed |= applyFailedShard(allocation, shardToFail, false);
}
// now, go over and elect a new primary if possible, not, from this code block on, if one is elected,
// routingNodes.hasUnassignedPrimaries() will potentially be false
for (MutableShardRouting shardEntry : routingNodes.unassigned()) {
if (shardEntry.primary()) {
MutableShardRouting candidate = allocation.routingNodes().activeReplica(shardEntry);
if (candidate != null) {
routingNodes.swapPrimaryFlag(shardEntry, candidate);
if (candidate.relocatingNodeId() != null) {
changed = true;
// its also relocating, make sure to move the other routing to primary
RoutingNode node = routingNodes.node(candidate.relocatingNodeId());
if (node != null) {
for (MutableShardRouting shardRouting : node) {
if (shardRouting.shardId().equals(candidate.shardId()) && !shardRouting.primary()) {
routingNodes.swapPrimaryFlag(shardRouting);
break;
}
}
}
}
}
}
}
return changed;
}
/**
* Applies the new nodes to the routing nodes and returns them (just the
* new nodes);
*/
private void applyNewNodes(RoutingAllocation allocation) {
final RoutingNodes routingNodes = allocation.routingNodes();
for (ObjectCursor<DiscoveryNode> cursor : allocation.nodes().dataNodes().values()) {
DiscoveryNode node = cursor.value;
if (!routingNodes.isKnown(node)) {
routingNodes.addNode(node);
}
}
}
private boolean deassociateDeadNodes(RoutingAllocation allocation) {
boolean changed = false;
for (RoutingNodes.RoutingNodesIterator it = allocation.routingNodes().nodes(); it.hasNext(); ) {
RoutingNode node = it.next();
if (allocation.nodes().dataNodes().containsKey(node.nodeId())) {
// its a live node, continue
continue;
}
changed = true;
// now, go over all the shards routing on the node, and fail them
for (MutableShardRouting shardRouting : node.copyShards()) {
applyFailedShard(allocation, shardRouting, false);
}
// its a dead node, remove it, note, its important to remove it *after* we apply failed shard
// since it relies on the fact that the RoutingNode exists in the list of nodes
it.remove();
}
return changed;
}
private boolean applyStartedShards(RoutingNodes routingNodes, Iterable<? extends ShardRouting> startedShardEntries) {
boolean dirty = false;
// apply shards might be called several times with the same shard, ignore it
for (ShardRouting startedShard : startedShardEntries) {
assert startedShard.state() == INITIALIZING;
// retrieve the relocating node id before calling startedShard().
String relocatingNodeId = null;
RoutingNodes.RoutingNodeIterator currentRoutingNode = routingNodes.routingNodeIter(startedShard.currentNodeId());
if (currentRoutingNode != null) {
for (MutableShardRouting shard : currentRoutingNode) {
if (shard.shardId().equals(startedShard.shardId())) {
relocatingNodeId = shard.relocatingNodeId();
if (!shard.started()) {
dirty = true;
routingNodes.started(shard);
}
break;
}
}
}
// startedShard is the current state of the shard (post relocation for example)
// this means that after relocation, the state will be started and the currentNodeId will be
// the node we relocated to
if (relocatingNodeId == null) {
continue;
}
RoutingNodes.RoutingNodeIterator sourceRoutingNode = routingNodes.routingNodeIter(relocatingNodeId);
if (sourceRoutingNode != null) {
while (sourceRoutingNode.hasNext()) {
MutableShardRouting shard = sourceRoutingNode.next();
if (shard.shardId().equals(startedShard.shardId())) {
if (shard.relocating()) {
dirty = true;
sourceRoutingNode.remove();
break;
}
}
}
}
}
return dirty;
}
/**
* Applies the relevant logic to handle a failed shard. Returns <tt>true</tt> if changes happened that
* require relocation.
*/
private boolean applyFailedShard(RoutingAllocation allocation, ShardRouting failedShard, boolean addToIgnoreList) {
// create a copy of the failed shard, since we assume we can change possible references to it without
// changing the state of failed shard
failedShard = new ImmutableShardRouting(failedShard);
IndexRoutingTable indexRoutingTable = allocation.routingTable().index(failedShard.index());
if (indexRoutingTable == null) {
return false;
}
RoutingNodes routingNodes = allocation.routingNodes();
boolean dirty = false;
if (failedShard.relocatingNodeId() != null) {
// the shard is relocating, either in initializing (recovery from another node) or relocating (moving to another node)
if (failedShard.state() == INITIALIZING) {
// the shard is initializing and recovering from another node
// first, we need to cancel the current node that is being initialized
RoutingNodes.RoutingNodeIterator initializingNode = routingNodes.routingNodeIter(failedShard.currentNodeId());
if (initializingNode != null) {
while (initializingNode.hasNext()) {
MutableShardRouting shardRouting = initializingNode.next();
if (shardRouting.equals(failedShard)) {
dirty = true;
initializingNode.remove();
if (addToIgnoreList) {
// make sure we ignore this shard on the relevant node
allocation.addIgnoreShardForNode(failedShard.shardId(), failedShard.currentNodeId());
}
break;
}
}
}
if (dirty) {
// now, find the node that we are relocating *from*, and cancel its relocation
RoutingNode relocatingFromNode = routingNodes.node(failedShard.relocatingNodeId());
if (relocatingFromNode != null) {
for (MutableShardRouting shardRouting : relocatingFromNode) {
if (shardRouting.shardId().equals(failedShard.shardId()) && shardRouting.relocating()) {
dirty = true;
routingNodes.cancelRelocation(shardRouting);
break;
}
}
}
} else {
logger.debug("failed shard {} not found in routingNodes, ignoring it", failedShard);
}
return dirty;
} else if (failedShard.state() == RELOCATING) {
// the shard is relocating, meaning its the source the shard is relocating from
// first, we need to cancel the current relocation from the current node
// now, find the node that we are recovering from, cancel the relocation, remove it from the node
// and add it to the unassigned shards list...
RoutingNodes.RoutingNodeIterator relocatingFromNode = routingNodes.routingNodeIter(failedShard.currentNodeId());
if (relocatingFromNode != null) {
while (relocatingFromNode.hasNext()) {
MutableShardRouting shardRouting = relocatingFromNode.next();
if (shardRouting.equals(failedShard)) {
dirty = true;
relocatingFromNode.remove();
if (addToIgnoreList) {
// make sure we ignore this shard on the relevant node
allocation.addIgnoreShardForNode(failedShard.shardId(), failedShard.currentNodeId());
}
routingNodes.unassigned().add(new MutableShardRouting(failedShard.index(), failedShard.id(),
null, failedShard.primary(), ShardRoutingState.UNASSIGNED, failedShard.version() + 1));
break;
}
}
}
if (dirty) {
// next, we need to find the target initializing shard that is recovering from, and remove it...
RoutingNodes.RoutingNodeIterator initializingNode = routingNodes.routingNodeIter(failedShard.relocatingNodeId());
if (initializingNode != null) {
while (initializingNode.hasNext()) {
MutableShardRouting shardRouting = initializingNode.next();
if (shardRouting.shardId().equals(failedShard.shardId()) && shardRouting.state() == INITIALIZING) {
dirty = true;
initializingNode.remove();
}
}
}
} else {
logger.debug("failed shard {} not found in routingNodes, ignoring it", failedShard);
}
} else {
throw new ElasticsearchIllegalStateException("illegal state for a failed shard, relocating node id is set, but state does not match: " + failedShard);
}
} else {
// the shard is not relocating, its either started, or initializing, just cancel it and move on...
RoutingNodes.RoutingNodeIterator node = routingNodes.routingNodeIter(failedShard.currentNodeId());
if (node != null) {
while (node.hasNext()) {
MutableShardRouting shardRouting = node.next();
if (shardRouting.equals(failedShard)) {
dirty = true;
if (addToIgnoreList) {
// make sure we ignore this shard on the relevant node
allocation.addIgnoreShardForNode(failedShard.shardId(), failedShard.currentNodeId());
}
node.remove();
// move all the shards matching the failed shard to the end of the unassigned list
// so we give a chance for other allocations and won't create poison failed allocations
// that can keep other shards from being allocated (because of limits applied on how many
// shards we can start per node)
List<MutableShardRouting> shardsToMove = Lists.newArrayList();
for (Iterator<MutableShardRouting> unassignedIt = routingNodes.unassigned().iterator(); unassignedIt.hasNext(); ) {
MutableShardRouting unassignedShardRouting = unassignedIt.next();
if (unassignedShardRouting.shardId().equals(failedShard.shardId())) {
unassignedIt.remove();
shardsToMove.add(unassignedShardRouting);
}
}
if (!shardsToMove.isEmpty()) {
routingNodes.unassigned().addAll(shardsToMove);
}
routingNodes.unassigned().add(new MutableShardRouting(failedShard.index(), failedShard.id(), null,
null, failedShard.restoreSource(), failedShard.primary(), ShardRoutingState.UNASSIGNED, failedShard.version() + 1));
break;
}
}
}
if (!dirty) {
logger.debug("failed shard {} not found in routingNodes, ignoring it", failedShard);
}
}
return dirty;
}
}