add another reroute method to do no assignment but still elect new primaries on failed nodes
This commit is contained in:
parent
25eba4b60d
commit
e5c1ab0ca6
|
@ -102,6 +102,32 @@ public class ShardsAllocation extends AbstractComponent {
|
|||
return new RoutingAllocation.Result(true, new RoutingTable.Builder().updateNodes(routingNodes).build().validateRaiseException(clusterState.metaData()), allocation.explanation());
|
||||
}
|
||||
|
||||
/**
|
||||
* 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) {
|
||||
RoutingNodes routingNodes = clusterState.routingNodes();
|
||||
RoutingAllocation allocation = new RoutingAllocation(routingNodes, clusterState.nodes());
|
||||
Iterable<DiscoveryNode> dataNodes = allocation.nodes().dataNodes().values();
|
||||
boolean changed = false;
|
||||
// first, clear from the shards any node id they used to belong to that is now dead
|
||||
changed |= deassociateDeadNodes(allocation.routingNodes(), dataNodes);
|
||||
|
||||
// create a sorted list of from nodes with least number of shards to the maximum ones
|
||||
applyNewNodes(allocation.routingNodes(), dataNodes);
|
||||
|
||||
// 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 |= electPrimaries(allocation.routingNodes());
|
||||
|
||||
if (!changed) {
|
||||
return new RoutingAllocation.Result(false, clusterState.routingTable(), allocation.explanation());
|
||||
}
|
||||
return new RoutingAllocation.Result(true, new RoutingTable.Builder().updateNodes(routingNodes).build().validateRaiseException(clusterState.metaData()), allocation.explanation());
|
||||
}
|
||||
|
||||
private boolean reroute(RoutingAllocation allocation) {
|
||||
Iterable<DiscoveryNode> dataNodes = allocation.nodes().dataNodes().values();
|
||||
|
||||
|
|
|
@ -0,0 +1,163 @@
|
|||
/*
|
||||
* Licensed to Elastic Search and Shay Banon under one
|
||||
* or more contributor license agreements. See the NOTICE file
|
||||
* distributed with this work for additional information
|
||||
* regarding copyright ownership. Elastic Search 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 org.elasticsearch.cluster.ClusterState;
|
||||
import org.elasticsearch.cluster.metadata.MetaData;
|
||||
import org.elasticsearch.cluster.routing.RoutingNode;
|
||||
import org.elasticsearch.cluster.routing.RoutingNodes;
|
||||
import org.elasticsearch.cluster.routing.RoutingTable;
|
||||
import org.elasticsearch.common.logging.ESLogger;
|
||||
import org.elasticsearch.common.logging.Loggers;
|
||||
import org.testng.annotations.Test;
|
||||
|
||||
import static org.elasticsearch.cluster.ClusterState.*;
|
||||
import static org.elasticsearch.cluster.metadata.IndexMetaData.*;
|
||||
import static org.elasticsearch.cluster.metadata.MetaData.*;
|
||||
import static org.elasticsearch.cluster.node.DiscoveryNodes.*;
|
||||
import static org.elasticsearch.cluster.routing.RoutingBuilders.*;
|
||||
import static org.elasticsearch.cluster.routing.ShardRoutingState.*;
|
||||
import static org.elasticsearch.cluster.routing.allocation.RoutingAllocationTests.*;
|
||||
import static org.elasticsearch.common.settings.ImmutableSettings.*;
|
||||
import static org.hamcrest.MatcherAssert.*;
|
||||
import static org.hamcrest.Matchers.*;
|
||||
|
||||
public class FailedNodeRoutingTests {
|
||||
|
||||
private final ESLogger logger = Loggers.getLogger(FailedNodeRoutingTests.class);
|
||||
|
||||
@Test public void simpleFailedNodeTest() {
|
||||
ShardsAllocation strategy = new ShardsAllocation(settingsBuilder().put("cluster.routing.allocation.allow_rebalance", ClusterRebalanceNodeAllocation.ClusterRebalanceType.ALWAYS.toString()).build());
|
||||
|
||||
MetaData metaData = newMetaDataBuilder()
|
||||
.put(newIndexMetaDataBuilder("test1").numberOfShards(1).numberOfReplicas(1))
|
||||
.put(newIndexMetaDataBuilder("test2").numberOfShards(1).numberOfReplicas(1))
|
||||
.build();
|
||||
|
||||
RoutingTable routingTable = routingTable()
|
||||
.add(indexRoutingTable("test1").initializeEmpty(metaData.index("test1")))
|
||||
.add(indexRoutingTable("test2").initializeEmpty(metaData.index("test2")))
|
||||
.build();
|
||||
|
||||
ClusterState clusterState = newClusterStateBuilder().metaData(metaData).routingTable(routingTable).build();
|
||||
|
||||
logger.info("start 4 nodes");
|
||||
clusterState = newClusterStateBuilder().state(clusterState).nodes(newNodesBuilder().put(newNode("node1")).put(newNode("node2")).put(newNode("node3")).put(newNode("node4"))).build();
|
||||
RoutingTable prevRoutingTable = routingTable;
|
||||
routingTable = strategy.reroute(clusterState).routingTable();
|
||||
clusterState = newClusterStateBuilder().state(clusterState).routingTable(routingTable).build();
|
||||
|
||||
logger.info("start all the primary shards, replicas will start initializing");
|
||||
RoutingNodes routingNodes = clusterState.routingNodes();
|
||||
prevRoutingTable = routingTable;
|
||||
routingTable = strategy.applyStartedShards(clusterState, routingNodes.shardsWithState(INITIALIZING)).routingTable();
|
||||
clusterState = newClusterStateBuilder().state(clusterState).routingTable(routingTable).build();
|
||||
routingNodes = clusterState.routingNodes();
|
||||
|
||||
logger.info("start the replica shards");
|
||||
routingNodes = clusterState.routingNodes();
|
||||
prevRoutingTable = routingTable;
|
||||
routingTable = strategy.applyStartedShards(clusterState, routingNodes.shardsWithState(INITIALIZING)).routingTable();
|
||||
clusterState = newClusterStateBuilder().state(clusterState).routingTable(routingTable).build();
|
||||
routingNodes = clusterState.routingNodes();
|
||||
|
||||
assertThat(routingNodes.node("node1").numberOfShardsWithState(STARTED), equalTo(1));
|
||||
assertThat(routingNodes.node("node2").numberOfShardsWithState(STARTED), equalTo(1));
|
||||
assertThat(routingNodes.node("node3").numberOfShardsWithState(STARTED), equalTo(1));
|
||||
assertThat(routingNodes.node("node4").numberOfShardsWithState(STARTED), equalTo(1));
|
||||
|
||||
|
||||
logger.info("remove 2 nodes where primaries are allocated, reroute");
|
||||
|
||||
clusterState = newClusterStateBuilder().state(clusterState).nodes(newNodesBuilder().putAll(clusterState.nodes())
|
||||
.remove(routingTable.index("test1").shard(0).primaryShard().currentNodeId())
|
||||
.remove(routingTable.index("test2").shard(0).primaryShard().currentNodeId())
|
||||
)
|
||||
.build();
|
||||
prevRoutingTable = routingTable;
|
||||
routingTable = strategy.reroute(clusterState).routingTable();
|
||||
clusterState = newClusterStateBuilder().state(clusterState).routingTable(routingTable).build();
|
||||
routingNodes = clusterState.routingNodes();
|
||||
|
||||
for (RoutingNode routingNode : routingNodes) {
|
||||
assertThat(routingNode.numberOfShardsWithState(STARTED), equalTo(1));
|
||||
assertThat(routingNode.numberOfShardsWithState(INITIALIZING), equalTo(1));
|
||||
}
|
||||
}
|
||||
|
||||
@Test public void simpleFailedNodeTestNoReassign() {
|
||||
ShardsAllocation strategy = new ShardsAllocation(settingsBuilder().put("cluster.routing.allocation.allow_rebalance", ClusterRebalanceNodeAllocation.ClusterRebalanceType.ALWAYS.toString()).build());
|
||||
|
||||
MetaData metaData = newMetaDataBuilder()
|
||||
.put(newIndexMetaDataBuilder("test1").numberOfShards(1).numberOfReplicas(1))
|
||||
.put(newIndexMetaDataBuilder("test2").numberOfShards(1).numberOfReplicas(1))
|
||||
.build();
|
||||
|
||||
RoutingTable routingTable = routingTable()
|
||||
.add(indexRoutingTable("test1").initializeEmpty(metaData.index("test1")))
|
||||
.add(indexRoutingTable("test2").initializeEmpty(metaData.index("test2")))
|
||||
.build();
|
||||
|
||||
ClusterState clusterState = newClusterStateBuilder().metaData(metaData).routingTable(routingTable).build();
|
||||
|
||||
logger.info("start 4 nodes");
|
||||
clusterState = newClusterStateBuilder().state(clusterState).nodes(newNodesBuilder().put(newNode("node1")).put(newNode("node2")).put(newNode("node3")).put(newNode("node4"))).build();
|
||||
RoutingTable prevRoutingTable = routingTable;
|
||||
routingTable = strategy.reroute(clusterState).routingTable();
|
||||
clusterState = newClusterStateBuilder().state(clusterState).routingTable(routingTable).build();
|
||||
|
||||
logger.info("start all the primary shards, replicas will start initializing");
|
||||
RoutingNodes routingNodes = clusterState.routingNodes();
|
||||
prevRoutingTable = routingTable;
|
||||
routingTable = strategy.applyStartedShards(clusterState, routingNodes.shardsWithState(INITIALIZING)).routingTable();
|
||||
clusterState = newClusterStateBuilder().state(clusterState).routingTable(routingTable).build();
|
||||
routingNodes = clusterState.routingNodes();
|
||||
|
||||
logger.info("start the replica shards");
|
||||
routingNodes = clusterState.routingNodes();
|
||||
prevRoutingTable = routingTable;
|
||||
routingTable = strategy.applyStartedShards(clusterState, routingNodes.shardsWithState(INITIALIZING)).routingTable();
|
||||
clusterState = newClusterStateBuilder().state(clusterState).routingTable(routingTable).build();
|
||||
routingNodes = clusterState.routingNodes();
|
||||
|
||||
assertThat(routingNodes.node("node1").numberOfShardsWithState(STARTED), equalTo(1));
|
||||
assertThat(routingNodes.node("node2").numberOfShardsWithState(STARTED), equalTo(1));
|
||||
assertThat(routingNodes.node("node3").numberOfShardsWithState(STARTED), equalTo(1));
|
||||
assertThat(routingNodes.node("node4").numberOfShardsWithState(STARTED), equalTo(1));
|
||||
|
||||
|
||||
logger.info("remove 2 nodes where primaries are allocated, reroute");
|
||||
|
||||
clusterState = newClusterStateBuilder().state(clusterState).nodes(newNodesBuilder().putAll(clusterState.nodes())
|
||||
.remove(routingTable.index("test1").shard(0).primaryShard().currentNodeId())
|
||||
.remove(routingTable.index("test2").shard(0).primaryShard().currentNodeId())
|
||||
)
|
||||
.build();
|
||||
prevRoutingTable = routingTable;
|
||||
routingTable = strategy.rerouteWithNoReassign(clusterState).routingTable();
|
||||
clusterState = newClusterStateBuilder().state(clusterState).routingTable(routingTable).build();
|
||||
routingNodes = clusterState.routingNodes();
|
||||
|
||||
for (RoutingNode routingNode : routingNodes) {
|
||||
assertThat(routingNode.numberOfShardsWithState(STARTED), equalTo(1));
|
||||
}
|
||||
assertThat(routingNodes.unassigned().size(), equalTo(2));
|
||||
}
|
||||
}
|
Loading…
Reference in New Issue