Skip to content

Decouple shard allocation awareness from search and get requests #45735

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Sep 4, 2019
Merged

Decouple shard allocation awareness from search and get requests #45735

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
71d49e9
Decouple shard allocation awareness from search and get requests
jimczi Aug 20, 2019
a66a58f
Merge branch 'master' into allocation_awareness_search
elasticmachine Aug 29, 2019
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 10 additions & 2 deletions docs/reference/migration/migrate_8_0/search.asciidoc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -12,16 +12,24 @@ The `/{index}/{type}/_termvectors`, `/{index}/{type}/{id}/_termvectors` and `/{i
[float]
==== Removal of queries

The `common` query was deprecated in 7.x and has been removed in 8.0.
The `common` query, deprecated in 7.x, has been removed in 8.0.
The same functionality can be achieved by the `match` query if the total number of hits is not tracked.

[float]
===== Removal of query parameters

The `cutoff_frequency` parameter was deprecated in 7.x and has been removed in 8.0 from `match` and `multi_match` queries.
The `cutoff_frequency` parameter, deprecated in 7.x, has been removed in 8.0 from `match` and `multi_match` queries.
The same functionality can be achieved without any configuration provided that the total number of hits is not tracked.

[float]
===== Removal of sort parameters

The `nested_filter` and `nested_path` options, deprecated in 6.x, have been removed in favor of the `nested` context.


[float]
===== Shard allocation awareness in Search and Get requests

{es} will no longer prefer using shards in the same location (with the same awareness attribute values) to process
`_search` and `_get` requests. Adaptive replica selection (activated by default in this version) will route requests
more efficiently using the service time of prior inter-node communications.
4 changes: 0 additions & 4 deletions docs/reference/modules/cluster/allocation_awareness.asciidoc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,10 +17,6 @@ The allocation awareness settings can be configured in
`elasticsearch.yml` and updated dynamically with the
<<cluster-update-settings,cluster-update-settings>> API.

{es} prefers using shards in the same location (with the same
awareness attribute values) to process search or GET requests. Using local
shards is usually faster than crossing rack or zone boundaries.

NOTE: The number of attribute values determines how many shard copies are
allocated in each location. If the number of nodes in each location is
unbalanced and there are a lot of replicas, replica shards might be left
Expand Down
88 changes: 0 additions & 88 deletions server/src/main/java/org/elasticsearch/cluster/routing/IndexShardRoutingTable.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -19,13 +19,11 @@

package org.elasticsearch.cluster.routing;

import org.elasticsearch.cluster.node.DiscoveryNode;
import org.elasticsearch.cluster.node.DiscoveryNodes;
import org.elasticsearch.common.Nullable;
import org.elasticsearch.common.Randomness;
import org.elasticsearch.common.io.stream.StreamInput;
import org.elasticsearch.common.io.stream.StreamOutput;
import org.elasticsearch.common.util.Maps;
import org.elasticsearch.common.util.set.Sets;
import org.elasticsearch.index.Index;
import org.elasticsearch.index.shard.ShardId;
Expand All @@ -44,8 +42,6 @@
import java.util.Optional;
import java.util.Set;

import static java.util.Collections.emptyMap;

/**
* {@link IndexShardRoutingTable} encapsulates all instances of a single shard.
* Each Elasticsearch index consists of multiple shards, each shard encapsulates
Expand All @@ -67,10 +63,6 @@ public class IndexShardRoutingTable implements Iterable<ShardRouting> {
final Set<String> allAllocationIds;
final boolean allShardsStarted;

private volatile Map<AttributesKey, AttributesRoutings> activeShardsByAttributes = emptyMap();
private volatile Map<AttributesKey, AttributesRoutings> initializingShardsByAttributes = emptyMap();
private final Object shardsByAttributeMutex = new Object();

/**
* The initializing list, including ones that are initializing on a target node because of relocation.
* If we can come up with a better variable name, it would be nice...
Expand Down Expand Up @@ -549,86 +541,6 @@ public boolean equals(Object obj) {
}
}

static class AttributesRoutings {

public final List<ShardRouting> withSameAttribute;
public final List<ShardRouting> withoutSameAttribute;
public final int totalSize;

AttributesRoutings(List<ShardRouting> withSameAttribute, List<ShardRouting> withoutSameAttribute) {
this.withSameAttribute = withSameAttribute;
this.withoutSameAttribute = withoutSameAttribute;
this.totalSize = withoutSameAttribute.size() + withSameAttribute.size();
}
}

private AttributesRoutings getActiveAttribute(AttributesKey key, DiscoveryNodes nodes) {
AttributesRoutings shardRoutings = activeShardsByAttributes.get(key);
if (shardRoutings == null) {
synchronized (shardsByAttributeMutex) {
ArrayList<ShardRouting> from = new ArrayList<>(activeShards);
List<ShardRouting> to = collectAttributeShards(key, nodes, from);

shardRoutings = new AttributesRoutings(to, Collections.unmodifiableList(from));
activeShardsByAttributes = Maps.copyMapWithAddedEntry(activeShardsByAttributes, key, shardRoutings);
}
}
return shardRoutings;
}

private AttributesRoutings getInitializingAttribute(AttributesKey key, DiscoveryNodes nodes) {
AttributesRoutings shardRoutings = initializingShardsByAttributes.get(key);
if (shardRoutings == null) {
synchronized (shardsByAttributeMutex) {
ArrayList<ShardRouting> from = new ArrayList<>(allInitializingShards);
List<ShardRouting> to = collectAttributeShards(key, nodes, from);
shardRoutings = new AttributesRoutings(to, Collections.unmodifiableList(from));
initializingShardsByAttributes =
Maps.copyMapWithAddedEntry(initializingShardsByAttributes, key, shardRoutings);
}
}
return shardRoutings;
}

private static List<ShardRouting> collectAttributeShards(AttributesKey key, DiscoveryNodes nodes, ArrayList<ShardRouting> from) {
final ArrayList<ShardRouting> to = new ArrayList<>();
for (final String attribute : key.attributes) {
final String localAttributeValue = nodes.getLocalNode().getAttributes().get(attribute);
if (localAttributeValue != null) {
for (Iterator<ShardRouting> iterator = from.iterator(); iterator.hasNext(); ) {
ShardRouting fromShard = iterator.next();
final DiscoveryNode discoveryNode = nodes.get(fromShard.currentNodeId());
if (discoveryNode == null) {
iterator.remove(); // node is not present anymore - ignore shard
} else if (localAttributeValue.equals(discoveryNode.getAttributes().get(attribute))) {
iterator.remove();
to.add(fromShard);
}
}
}
}
return Collections.unmodifiableList(to);
}

public ShardIterator preferAttributesActiveInitializingShardsIt(List<String> attributes, DiscoveryNodes nodes) {
return preferAttributesActiveInitializingShardsIt(attributes, nodes, shuffler.nextSeed());
}

public ShardIterator preferAttributesActiveInitializingShardsIt(List<String> attributes, DiscoveryNodes nodes, int seed) {
AttributesKey key = new AttributesKey(attributes);
AttributesRoutings activeRoutings = getActiveAttribute(key, nodes);
AttributesRoutings initializingRoutings = getInitializingAttribute(key, nodes);

// we now randomize, once between the ones that have the same attributes, and once for the ones that don't
// we don't want to mix between the two!
ArrayList<ShardRouting> ordered = new ArrayList<>(activeRoutings.totalSize + initializingRoutings.totalSize);
ordered.addAll(shuffler.shuffle(activeRoutings.withSameAttribute, seed));
ordered.addAll(shuffler.shuffle(activeRoutings.withoutSameAttribute, seed));
ordered.addAll(shuffler.shuffle(initializingRoutings.withSameAttribute, seed));
ordered.addAll(shuffler.shuffle(initializingRoutings.withoutSameAttribute, seed));
return new PlainShardIterator(shardId, ordered);
}

public ShardRouting primaryShard() {
return primary;
}
Expand Down
26 changes: 4 additions & 22 deletions server/src/main/java/org/elasticsearch/cluster/routing/OperationRouting.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -22,7 +22,6 @@
import org.elasticsearch.cluster.ClusterState;
import org.elasticsearch.cluster.metadata.IndexMetaData;
import org.elasticsearch.cluster.node.DiscoveryNodes;
import org.elasticsearch.cluster.routing.allocation.decider.AwarenessAllocationDecider;
import org.elasticsearch.common.Nullable;
import org.elasticsearch.common.Strings;
import org.elasticsearch.common.settings.ClusterSettings;
Expand All @@ -36,7 +35,6 @@
import java.util.Arrays;
import java.util.Collections;
import java.util.HashSet;
import java.util.List;
import java.util.Map;
import java.util.Set;
import java.util.stream.Collectors;
Expand All @@ -47,25 +45,17 @@ public class OperationRouting {
Setting.boolSetting("cluster.routing.use_adaptive_replica_selection", true,
Setting.Property.Dynamic, Setting.Property.NodeScope);

private List<String> awarenessAttributes;
private boolean useAdaptiveReplicaSelection;

public OperationRouting(Settings settings, ClusterSettings clusterSettings) {
this.awarenessAttributes = AwarenessAllocationDecider.CLUSTER_ROUTING_ALLOCATION_AWARENESS_ATTRIBUTE_SETTING.get(settings);
this.useAdaptiveReplicaSelection = USE_ADAPTIVE_REPLICA_SELECTION_SETTING.get(settings);
clusterSettings.addSettingsUpdateConsumer(AwarenessAllocationDecider.CLUSTER_ROUTING_ALLOCATION_AWARENESS_ATTRIBUTE_SETTING,
this::setAwarenessAttributes);
clusterSettings.addSettingsUpdateConsumer(USE_ADAPTIVE_REPLICA_SELECTION_SETTING, this::setUseAdaptiveReplicaSelection);
}

void setUseAdaptiveReplicaSelection(boolean useAdaptiveReplicaSelection) {
this.useAdaptiveReplicaSelection = useAdaptiveReplicaSelection;
}

private void setAwarenessAttributes(List<String> awarenessAttributes) {
this.awarenessAttributes = awarenessAttributes;
}

public ShardIterator indexShards(ClusterState clusterState, String index, String id, @Nullable String routing) {
return shards(clusterState, index, id, routing).shardsIt();
}
Expand Down Expand Up @@ -194,23 +184,15 @@ private ShardIterator preferenceActiveShardIterator(IndexShardRoutingTable index
}
// if not, then use it as the index
int routingHash = 31 * Murmur3HashFunction.hash(preference) + indexShard.shardId.hashCode();
if (awarenessAttributes.isEmpty()) {
return indexShard.activeInitializingShardsIt(routingHash);
} else {
return indexShard.preferAttributesActiveInitializingShardsIt(awarenessAttributes, nodes, routingHash);
}
return indexShard.activeInitializingShardsIt(routingHash);
}

private ShardIterator shardRoutings(IndexShardRoutingTable indexShard, DiscoveryNodes nodes,
@Nullable ResponseCollectorService collectorService, @Nullable Map<String, Long> nodeCounts) {
if (awarenessAttributes.isEmpty()) {
if (useAdaptiveReplicaSelection) {
return indexShard.activeInitializingShardsRankedIt(collectorService, nodeCounts);
} else {
return indexShard.activeInitializingShardsRandomIt();
}
if (useAdaptiveReplicaSelection) {
return indexShard.activeInitializingShardsRankedIt(collectorService, nodeCounts);
} else {
return indexShard.preferAttributesActiveInitializingShardsIt(awarenessAttributes, nodes);
return indexShard.activeInitializingShardsRandomIt();
}
}

Expand Down
55 changes: 0 additions & 55 deletions server/src/test/java/org/elasticsearch/cluster/structure/RoutingIteratorTests.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -41,10 +41,8 @@
import org.elasticsearch.common.settings.Settings;
import org.elasticsearch.index.shard.ShardId;

import java.util.Arrays;
import java.util.Collections;
import java.util.Iterator;
import java.util.Map;

import static java.util.Collections.singletonMap;
import static org.hamcrest.Matchers.anyOf;
Expand Down Expand Up @@ -224,59 +222,6 @@ public void testRandomRouting() {
assertThat(shardRouting1, sameInstance(shardRouting3));
}

public void testAttributePreferenceRouting() {
Settings.Builder settings = Settings.builder()
.put("cluster.routing.allocation.node_concurrent_recoveries", 10)
.put(ClusterRebalanceAllocationDecider.CLUSTER_ROUTING_ALLOCATION_ALLOW_REBALANCE_SETTING.getKey(), "always");
if (randomBoolean()) {
settings.put("cluster.routing.allocation.awareness.attributes", " rack_id, zone ");
} else {
settings.putList("cluster.routing.allocation.awareness.attributes", "rack_id", "zone");
}

AllocationService strategy = createAllocationService(settings.build());

MetaData metaData = MetaData.builder()
.put(IndexMetaData.builder("test").settings(settings(Version.CURRENT)).numberOfShards(1).numberOfReplicas(1))
.build();

RoutingTable routingTable = RoutingTable.builder()
.addAsNew(metaData.index("test"))
.build();

ClusterState clusterState = ClusterState.builder(ClusterName.CLUSTER_NAME_SETTING
.getDefault(Settings.EMPTY)).metaData(metaData).routingTable(routingTable).build();

clusterState = ClusterState.builder(clusterState).nodes(DiscoveryNodes.builder()
.add(newNode("node1", Map.of("rack_id", "rack_1", "zone", "zone1")))
.add(newNode("node2", Map.of("rack_id", "rack_2", "zone", "zone2")))
.localNodeId("node1")
).build();
clusterState = strategy.reroute(clusterState, "reroute");

clusterState = startInitializingShardsAndReroute(strategy, clusterState);
clusterState = startInitializingShardsAndReroute(strategy, clusterState);

// after all are started, check routing iteration
ShardIterator shardIterator = clusterState.routingTable().index("test").shard(0)
.preferAttributesActiveInitializingShardsIt(Arrays.asList("rack_id"), clusterState.nodes());
ShardRouting shardRouting = shardIterator.nextOrNull();
assertThat(shardRouting, notNullValue());
assertThat(shardRouting.currentNodeId(), equalTo("node1"));
shardRouting = shardIterator.nextOrNull();
assertThat(shardRouting, notNullValue());
assertThat(shardRouting.currentNodeId(), equalTo("node2"));

shardIterator = clusterState.routingTable().index("test").shard(0)
.preferAttributesActiveInitializingShardsIt(Arrays.asList("rack_id"), clusterState.nodes());
shardRouting = shardIterator.nextOrNull();
assertThat(shardRouting, notNullValue());
assertThat(shardRouting.currentNodeId(), equalTo("node1"));
shardRouting = shardIterator.nextOrNull();
assertThat(shardRouting, notNullValue());
assertThat(shardRouting.currentNodeId(), equalTo("node2"));
}

public void testNodeSelectorRouting(){
AllocationService strategy = createAllocationService(Settings.builder()
.put("cluster.routing.allocation.node_concurrent_recoveries", 10)
Expand Down