honeymoose
/
OpenSearch
mirror of https://github.com/honeymoose/OpenSearch.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Improve NodeVersionAllocationDecider messages (#29356) 

Since #26542 the NodeVersionAllocationDecider tries to explain its NO decisions
as follows:

    ... may not support codecs or postings formats for a newer Lucene version

However, this message often appears during a rolling upgrade, and experience
has shown that it seems to cause more confusion and worry than it needs to.
This change fixes that by removing the explanation again, reducing the message
to a statement of fact about the respective nodes' versions.

Additionally, the same wording was used for version incompatibilities when
allocating a primary (vs its previous location) and a replica (vs its primary).
This change separates these two cases so they can have separate, clearer
wording.

Fixes #29228
This commit is contained in:
David Turner 2018-04-05 15:13:48 +01:00 committed by GitHub
parent d813a05b9f
commit fb1aba9389
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 111 additions and 13 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

37
server/src/main/java/org/elasticsearch/cluster/routing/allocation/decider/NodeVersionAllocationDecider.java
View File

@ -55,13 +55,13 @@ public class NodeVersionAllocationDecider extends AllocationDecider {
}
} else {
// relocating primary, only migrate to newer host
return isVersionCompatible(allocation.routingNodes(), shardRouting.currentNodeId(), node, allocation);
return isVersionCompatibleRelocatePrimary(allocation.routingNodes(), shardRouting.currentNodeId(), node, allocation);
}
} else {
final ShardRouting primary = allocation.routingNodes().activePrimary(shardRouting.shardId());
// check that active primary has a newer version so that peer recovery works
if (primary != null) {
return isVersionCompatible(allocation.routingNodes(), primary.currentNodeId(), node, allocation);
return isVersionCompatibleAllocatingReplica(allocation.routingNodes(), primary.currentNodeId(), node, allocation);
} else {
// ReplicaAfterPrimaryActiveAllocationDecider should prevent this case from occurring
return allocation.decision(Decision.YES, NAME, "no active primary shard yet");
@ -69,30 +69,45 @@ public class NodeVersionAllocationDecider extends AllocationDecider {
}
}
private Decision isVersionCompatible(final RoutingNodes routingNodes, final String sourceNodeId, final RoutingNode target,
RoutingAllocation allocation) {
private Decision isVersionCompatibleRelocatePrimary(final RoutingNodes routingNodes, final String sourceNodeId,
final RoutingNode target, final RoutingAllocation allocation) {
final RoutingNode source = routingNodes.node(sourceNodeId);
if (target.node().getVersion().onOrAfter(source.node().getVersion())) {
return allocation.decision(Decision.YES, NAME,
"can relocate primary shard from a node with version [%s] to a node with equal-or-newer version [%s]",
source.node().getVersion(), target.node().getVersion());
} else {
return allocation.decision(Decision.NO, NAME,
"cannot relocate primary shard from a node with version [%s] to a node with older version [%s]",
source.node().getVersion(), target.node().getVersion());
}
}
private Decision isVersionCompatibleAllocatingReplica(final RoutingNodes routingNodes, final String sourceNodeId,
final RoutingNode target, final RoutingAllocation allocation) {
final RoutingNode source = routingNodes.node(sourceNodeId);
if (target.node().getVersion().onOrAfter(source.node().getVersion())) {
/* we can allocate if we can recover from a node that is younger or on the same version
* if the primary is already running on a newer version that won't work due to possible
* differences in the lucene index format etc.*/
return allocation.decision(Decision.YES, NAME, "target node version [%s] is the same or newer than source node version [%s]",
return allocation.decision(Decision.YES, NAME,
"can allocate replica shard to a node with version [%s] since this is equal-or-newer than the primary version [%s]",
target.node().getVersion(), source.node().getVersion());
} else {
return allocation.decision(Decision.NO, NAME, "target node version [%s] is older than the source node version [%s] and may " +
"not support codecs or postings formats for a newer Lucene version",
return allocation.decision(Decision.NO, NAME,
"cannot allocate replica shard to a node with version [%s] since this is older than the primary version [%s]",
target.node().getVersion(), source.node().getVersion());
}
}
private Decision isVersionCompatible(SnapshotRecoverySource recoverySource, final RoutingNode target, RoutingAllocation allocation) {
private Decision isVersionCompatible(SnapshotRecoverySource recoverySource, final RoutingNode target,
final RoutingAllocation allocation) {
if (target.node().getVersion().onOrAfter(recoverySource.version())) {
/* we can allocate if we can restore from a snapshot that is older or on the same version */
return allocation.decision(Decision.YES, NAME, "target node version [%s] is the same or newer than snapshot version [%s]",
return allocation.decision(Decision.YES, NAME, "node version [%s] is the same or newer than snapshot version [%s]",
target.node().getVersion(), recoverySource.version());
} else {
return allocation.decision(Decision.NO, NAME, "target node version [%s] is older than the snapshot version [%s] and may " +
"not support codecs or postings formats for a newer Lucene version",
return allocation.decision(Decision.NO, NAME, "node version [%s] is older than the snapshot version [%s]",
target.node().getVersion(), recoverySource.version());
}
}

83
server/src/test/java/org/elasticsearch/cluster/routing/allocation/NodeVersionAllocationDeciderTests.java
View File

@ -33,15 +33,19 @@ import org.elasticsearch.cluster.routing.AllocationId;
import org.elasticsearch.cluster.routing.IndexRoutingTable;
import org.elasticsearch.cluster.routing.IndexShardRoutingTable;
import org.elasticsearch.cluster.routing.RecoverySource.SnapshotRecoverySource;
import org.elasticsearch.cluster.routing.RoutingChangesObserver;
import org.elasticsearch.cluster.routing.RoutingNode;
import org.elasticsearch.cluster.routing.RoutingNodes;
import org.elasticsearch.cluster.routing.RoutingTable;
import org.elasticsearch.cluster.routing.ShardRouting;
import org.elasticsearch.cluster.routing.ShardRoutingHelper;
import org.elasticsearch.cluster.routing.ShardRoutingState;
import org.elasticsearch.cluster.routing.TestShardRouting;
import org.elasticsearch.cluster.routing.allocation.allocator.BalancedShardsAllocator;
import org.elasticsearch.cluster.routing.allocation.command.AllocationCommands;
import org.elasticsearch.cluster.routing.allocation.decider.AllocationDeciders;
import org.elasticsearch.cluster.routing.allocation.decider.ClusterRebalanceAllocationDecider;
import org.elasticsearch.cluster.routing.allocation.decider.Decision;
import org.elasticsearch.cluster.routing.allocation.decider.NodeVersionAllocationDecider;
import org.elasticsearch.cluster.routing.allocation.decider.ReplicaAfterPrimaryActiveAllocationDecider;
import org.elasticsearch.common.UUIDs;
@ -68,6 +72,7 @@ import static org.elasticsearch.test.VersionUtils.randomVersion;
import static org.hamcrest.Matchers.equalTo;
import static org.hamcrest.Matchers.notNullValue;
import static org.hamcrest.Matchers.nullValue;
import static org.hamcrest.core.Is.is;
public class NodeVersionAllocationDeciderTests extends ESAllocationTestCase {
@ -428,4 +433,82 @@ public class NodeVersionAllocationDeciderTests extends ESAllocationTestCase {
}
}
public void testMessages() {
MetaData metaData = MetaData.builder()
.put(IndexMetaData.builder("test").settings(settings(Version.CURRENT)).numberOfShards(1).numberOfReplicas(1))
.build();
RoutingTable initialRoutingTable = RoutingTable.builder()
.addAsNew(metaData.index("test"))
.build();
RoutingNode newNode = new RoutingNode("newNode", newNode("newNode", Version.CURRENT));
RoutingNode oldNode = new RoutingNode("oldNode", newNode("oldNode", VersionUtils.getPreviousVersion()));
final ClusterName clusterName = ClusterName.CLUSTER_NAME_SETTING.getDefault(Settings.EMPTY);
ClusterState clusterState = ClusterState.builder(clusterName).metaData(metaData).routingTable(initialRoutingTable)
.nodes(DiscoveryNodes.builder().add(newNode.node()).add(oldNode.node())).build();
final ShardId shardId = clusterState.routingTable().index("test").shard(0).getShardId();
final ShardRouting primaryShard = clusterState.routingTable().shardRoutingTable(shardId).primaryShard();
final ShardRouting replicaShard = clusterState.routingTable().shardRoutingTable(shardId).replicaShards().get(0);
RoutingAllocation routingAllocation = new RoutingAllocation(null, clusterState.getRoutingNodes(), clusterState, null, 0);
routingAllocation.debugDecision(true);
final NodeVersionAllocationDecider allocationDecider = new NodeVersionAllocationDecider(Settings.EMPTY);
Decision decision = allocationDecider.canAllocate(primaryShard, newNode, routingAllocation);
assertThat(decision.type(), is(Decision.Type.YES));
assertThat(decision.getExplanation(), is("the primary shard is new or already existed on the node"));
decision = allocationDecider.canAllocate(ShardRoutingHelper.initialize(primaryShard, "oldNode"), newNode, routingAllocation);
assertThat(decision.type(), is(Decision.Type.YES));
assertThat(decision.getExplanation(), is("can relocate primary shard from a node with version [" +
oldNode.node().getVersion() + "] to a node with equal-or-newer version [" + newNode.node().getVersion() + "]"));
decision = allocationDecider.canAllocate(ShardRoutingHelper.initialize(primaryShard, "newNode"), oldNode, routingAllocation);
assertThat(decision.type(), is(Decision.Type.NO));
assertThat(decision.getExplanation(), is("cannot relocate primary shard from a node with version [" +
newNode.node().getVersion() + "] to a node with older version [" + oldNode.node().getVersion() + "]"));
final SnapshotRecoverySource newVersionSnapshot = new SnapshotRecoverySource(
new Snapshot("rep1", new SnapshotId("snp1", UUIDs.randomBase64UUID())), newNode.node().getVersion(), "test");
final SnapshotRecoverySource oldVersionSnapshot = new SnapshotRecoverySource(
new Snapshot("rep1", new SnapshotId("snp1", UUIDs.randomBase64UUID())), oldNode.node().getVersion(), "test");
decision = allocationDecider.canAllocate(ShardRoutingHelper.newWithRestoreSource(primaryShard, newVersionSnapshot),
oldNode, routingAllocation);
assertThat(decision.type(), is(Decision.Type.NO));
assertThat(decision.getExplanation(), is("node version [" +
oldNode.node().getVersion() + "] is older than the snapshot version [" + newNode.node().getVersion() + "]"));
decision = allocationDecider.canAllocate(ShardRoutingHelper.newWithRestoreSource(primaryShard, oldVersionSnapshot),
newNode, routingAllocation);
assertThat(decision.type(), is(Decision.Type.YES));
assertThat(decision.getExplanation(), is("node version [" +
newNode.node().getVersion() + "] is the same or newer than snapshot version [" + oldNode.node().getVersion() + "]"));
final RoutingChangesObserver routingChangesObserver = new RoutingChangesObserver.AbstractRoutingChangesObserver();
final RoutingNodes routingNodes = new RoutingNodes(clusterState, false);
final ShardRouting startedPrimary = routingNodes.startShard(logger, routingNodes.initializeShard(primaryShard, "newNode", null, 0,
routingChangesObserver), routingChangesObserver);
routingAllocation = new RoutingAllocation(null, routingNodes, clusterState, null, 0);
routingAllocation.debugDecision(true);
decision = allocationDecider.canAllocate(replicaShard, oldNode, routingAllocation);
assertThat(decision.type(), is(Decision.Type.NO));
assertThat(decision.getExplanation(), is("cannot allocate replica shard to a node with version [" +
oldNode.node().getVersion() + "] since this is older than the primary version [" + newNode.node().getVersion() + "]"));
routingNodes.startShard(logger, routingNodes.relocateShard(startedPrimary, "oldNode", 0, routingChangesObserver).v2(),
routingChangesObserver);
routingAllocation = new RoutingAllocation(null, routingNodes, clusterState, null, 0);
routingAllocation.debugDecision(true);
decision = allocationDecider.canAllocate(replicaShard, newNode, routingAllocation);
assertThat(decision.type(), is(Decision.Type.YES));
assertThat(decision.getExplanation(), is("can allocate replica shard to a node with version [" +
newNode.node().getVersion() + "] since this is equal-or-newer than the primary version [" + oldNode.node().getVersion() + "]"));
}
}