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

Relax fault detector in some disruption tests (#38101) 

Today we use `AbstractDisruptionTestCase` to test the behaviour of things like
master elections in the presence of cluster disruptions. These tests have
rather enthusiastic fault detection settings, detecting a fault if a single
ping fails, with a one-second timeout. Furthermore there are some tests that
assert the identity of the master remains unchanged during some disruption, and
these assertions fail rather often thanks to the overly sensitive fault
detector.

However in a number of these tests the fault detector need not be this
sensitive. This commit moves some such tests into their own test suite and uses
more sensible fault-detection settings to avoid the kind of master instability
that is causing CI failures.

Closes #37699
This commit is contained in:
David Turner 2019-02-01 08:10:49 +00:00 committed by GitHub
parent c02cd3e2fd
commit 23f00e3676
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 170 additions and 71 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

71
server/src/test/java/org/elasticsearch/discovery/MasterDisruptionIT.java
View File

@ -30,7 +30,6 @@ import org.elasticsearch.cluster.metadata.IndexMetaData;
import org.elasticsearch.cluster.node.DiscoveryNode;
import org.elasticsearch.cluster.service.ClusterService;
import org.elasticsearch.common.Priority;
import org.elasticsearch.common.Strings;
import org.elasticsearch.common.collect.Tuple;
import org.elasticsearch.common.settings.Settings;
import org.elasticsearch.common.unit.TimeValue;
@ -71,76 +70,6 @@ import static org.hamcrest.Matchers.nullValue;
@ESIntegTestCase.ClusterScope(scope = ESIntegTestCase.Scope.TEST, numDataNodes = 0, transportClientRatio = 0)
public class MasterDisruptionIT extends AbstractDisruptionTestCase {
/**
* Test that no split brain occurs under partial network partition. See https://github.com/elastic/elasticsearch/issues/2488
*/
public void testFailWithMinimumMasterNodesConfigured() throws Exception {
List<String> nodes = startCluster(3);
// Figure out what is the elected master node
final String masterNode = internalCluster().getMasterName();
logger.info("---> legit elected master node={}", masterNode);
// Pick a node that isn't the elected master.
Set<String> nonMasters = new HashSet<>(nodes);
nonMasters.remove(masterNode);
final String unluckyNode = randomFrom(nonMasters.toArray(Strings.EMPTY_ARRAY));
// Simulate a network issue between the unlucky node and elected master node in both directions.
NetworkDisruption networkDisconnect = new NetworkDisruption(
new NetworkDisruption.TwoPartitions(masterNode, unluckyNode),
new NetworkDisruption.NetworkDisconnect());
setDisruptionScheme(networkDisconnect);
networkDisconnect.startDisrupting();
// Wait until elected master has removed that the unlucky node...
ensureStableCluster(2, masterNode);
// The unlucky node must report *no* master node, since it can't connect to master and in fact it should
// continuously ping until network failures have been resolved. However
// It may a take a bit before the node detects it has been cut off from the elected master
assertNoMaster(unluckyNode);
networkDisconnect.stopDisrupting();
// Wait until the master node sees all 3 nodes again.
ensureStableCluster(3);
// The elected master shouldn't have changed, since the unlucky node never could have elected himself as
// master since m_m_n of 2 could never be satisfied.
assertMaster(masterNode, nodes);
}
/**
* Verify that nodes fault detection works after master (re) election
*/
public void testNodesFDAfterMasterReelection() throws Exception {
startCluster(4);
logger.info("--> stopping current master");
internalCluster().stopCurrentMasterNode();
ensureStableCluster(3);
String master = internalCluster().getMasterName();
String nonMaster = null;
for (String node : internalCluster().getNodeNames()) {
if (!node.equals(master)) {
nonMaster = node;
}
}
logger.info("--> isolating [{}]", nonMaster);
NetworkDisruption.TwoPartitions partitions = isolateNode(nonMaster);
NetworkDisruption networkDisruption = addRandomDisruptionType(partitions);
networkDisruption.startDisrupting();
logger.info("--> waiting for master to remove it");
ensureStableCluster(2, master);
}
/**
* Tests that emulates a frozen elected master node that unfreezes and pushes his cluster state to other nodes
* that already are following another elected master node. These nodes should reject this cluster state and prevent

170
server/src/test/java/org/elasticsearch/discovery/StableMasterDisruptionIT.java Normal file
View File

@ -0,0 +1,170 @@
/*
* 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.discovery;
import org.elasticsearch.action.admin.cluster.state.ClusterStateRequest;
import org.elasticsearch.cluster.coordination.FollowersChecker;
import org.elasticsearch.cluster.coordination.LeaderChecker;
import org.elasticsearch.common.Strings;
import org.elasticsearch.common.settings.Settings;
import org.elasticsearch.common.util.set.Sets;
import org.elasticsearch.plugins.Plugin;
import org.elasticsearch.test.ESIntegTestCase;
import org.elasticsearch.test.disruption.NetworkDisruption;
import org.elasticsearch.test.disruption.NetworkDisruption.NetworkDisconnect;
import org.elasticsearch.test.disruption.NetworkDisruption.NetworkUnresponsive;
import org.elasticsearch.test.junit.annotations.TestLogging;
import org.elasticsearch.test.transport.MockTransportService.TestPlugin;
import java.util.Arrays;
import java.util.Collection;
import java.util.Collections;
import java.util.HashSet;
import java.util.List;
import java.util.Set;
import java.util.stream.Collectors;
import java.util.stream.Stream;
import static java.util.Collections.singleton;
import static org.hamcrest.Matchers.equalTo;
/**
* Tests relating to the loss of the master, but which work with the default fault detection settings which are rather lenient and will
* not detect a master failure too quickly.
*/
@TestLogging("_root:DEBUG,org.elasticsearch.cluster.service:TRACE")
@ESIntegTestCase.ClusterScope(scope = ESIntegTestCase.Scope.TEST, numDataNodes = 0, transportClientRatio = 0)
public class StableMasterDisruptionIT extends ESIntegTestCase {
@Override
protected Collection<Class<? extends Plugin>> nodePlugins() {
return Collections.singletonList(TestPlugin.class);
}
/**
* Test that no split brain occurs under partial network partition. See https://github.com/elastic/elasticsearch/issues/2488
*/
public void testFailWithMinimumMasterNodesConfigured() throws Exception {
List<String> nodes = internalCluster().startNodes(3);
ensureStableCluster(3);
// Figure out what is the elected master node
final String masterNode = internalCluster().getMasterName();
logger.info("---> legit elected master node={}", masterNode);
// Pick a node that isn't the elected master.
Set<String> nonMasters = new HashSet<>(nodes);
nonMasters.remove(masterNode);
final String unluckyNode = randomFrom(nonMasters.toArray(Strings.EMPTY_ARRAY));
// Simulate a network issue between the unlucky node and elected master node in both directions.
NetworkDisruption networkDisconnect = new NetworkDisruption(
new NetworkDisruption.TwoPartitions(masterNode, unluckyNode),
new NetworkDisruption.NetworkDisconnect());
setDisruptionScheme(networkDisconnect);
networkDisconnect.startDisrupting();
// Wait until elected master has removed that the unlucky node...
ensureStableCluster(2, masterNode);
// The unlucky node must report *no* master node, since it can't connect to master and in fact it should
// continuously ping until network failures have been resolved. However
// It may a take a bit before the node detects it has been cut off from the elected master
assertBusy(() -> assertNull(client(unluckyNode).admin().cluster().state(
new ClusterStateRequest().local(true)).get().getState().nodes().getMasterNode()));
networkDisconnect.stopDisrupting();
// Wait until the master node sees all 3 nodes again.
ensureStableCluster(3);
// The elected master shouldn't have changed, since the unlucky node never could have elected itself as master
assertThat(internalCluster().getMasterName(), equalTo(masterNode));
}
/**
* Verify that nodes fault detection works after master (re) election
*/
public void testFollowerCheckerDetectsUnresponsiveNodeAfterMasterReelection() throws Exception {
internalCluster().startNodes(4,
Settings.builder()
.put(LeaderChecker.LEADER_CHECK_TIMEOUT_SETTING.getKey(), "1s")
.put(LeaderChecker.LEADER_CHECK_RETRY_COUNT_SETTING.getKey(), "10")
.put(FollowersChecker.FOLLOWER_CHECK_TIMEOUT_SETTING.getKey(), "1s")
.put(FollowersChecker.FOLLOWER_CHECK_RETRY_COUNT_SETTING.getKey(), 1).build());
ensureStableCluster(4);
logger.info("--> stopping current master");
internalCluster().stopCurrentMasterNode();
ensureStableCluster(3);
final String master = internalCluster().getMasterName();
final List<String> nonMasters = Arrays.stream(internalCluster().getNodeNames()).filter(n -> master.equals(n) == false)
.collect(Collectors.toList());
final String isolatedNode = randomFrom(nonMasters);
final String otherNode = nonMasters.get(nonMasters.get(0).equals(isolatedNode) ? 1 : 0);
logger.info("--> isolating [{}]", isolatedNode);
final NetworkDisruption networkDisruption = new NetworkDisruption(new NetworkDisruption.TwoPartitions(
singleton(isolatedNode), Sets.newHashSet(master, otherNode)), new NetworkUnresponsive());
setDisruptionScheme(networkDisruption);
networkDisruption.startDisrupting();
logger.info("--> waiting for master to remove it");
ensureStableCluster(2, master);
networkDisruption.stopDisrupting();
ensureStableCluster(3);
}
/**
* Verify that nodes fault detection works after master (re) election
*/
public void testFollowerCheckerDetectsDisconnectedNodeAfterMasterReelection() throws Exception {
internalCluster().startNodes(4);
ensureStableCluster(4);
logger.info("--> stopping current master");
internalCluster().stopCurrentMasterNode();
ensureStableCluster(3);
final String master = internalCluster().getMasterName();
final List<String> nonMasters = Arrays.stream(internalCluster().getNodeNames()).filter(n -> master.equals(n) == false)
.collect(Collectors.toList());
final String isolatedNode = randomFrom(nonMasters);
final String otherNode = nonMasters.get(nonMasters.get(0).equals(isolatedNode) ? 1 : 0);
logger.info("--> isolating [{}]", isolatedNode);
final NetworkDisruption networkDisruption = new NetworkDisruption(new NetworkDisruption.TwoPartitions(
singleton(isolatedNode), Stream.of(master, otherNode).collect(Collectors.toSet())), new NetworkDisconnect());
setDisruptionScheme(networkDisruption);
networkDisruption.startDisrupting();
logger.info("--> waiting for master to remove it");
ensureStableCluster(2, master);
networkDisruption.stopDisrupting();
ensureStableCluster(3);
}
}