Skip to content

Commit

Permalink
CCR/TEST: Limit indexing docs in FollowerFailOverIT (#35228)
Browse files Browse the repository at this point in the history 
The suite FollowerFailOverIT is failing because some documents are not
replicated to the follower. Maybe the FollowTask is not working as
expected or the background indexers eat all resources while the follower
cluster is trying to reform after a failover; then CI is not fast enough
to replicate all the indexed docs within 60 seconds (sometimes I see 80k
docs on the leader).

This commit limits the number of documents to be indexed into the leader
index by the background threads so that we can eliminate the latter
case. This change also replaces a docCount assertion with a docIds
assertion so we can have more information if these tests fail again.

Relates #33337
  • Loading branch information
@dnhatn
dnhatn committed Nov 3, 2018
1 parent 18bcf27 commit 96af589
Show file tree
Hide file tree
Showing 3 changed files with 130 additions and 41 deletions.
79 changes: 69 additions & 10 deletions x-pack/plugin/ccr/src/test/java/org/elasticsearch/xpack/CcrIntegTestCase.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -15,6 +15,7 @@
import org.elasticsearch.action.admin.cluster.settings.ClusterUpdateSettingsRequest;
import org.elasticsearch.action.admin.indices.get.GetIndexResponse;
import org.elasticsearch.action.admin.indices.refresh.RefreshResponse;
import org.elasticsearch.action.admin.indices.stats.ShardStats;
import org.elasticsearch.action.search.SearchRequest;
import org.elasticsearch.action.search.SearchResponse;
import org.elasticsearch.analysis.common.CommonAnalysisPlugin;
Expand All @@ -25,9 +26,11 @@
import org.elasticsearch.cluster.health.ClusterHealthStatus;
import org.elasticsearch.cluster.metadata.IndexMetaData;
import org.elasticsearch.cluster.metadata.MetaData;
import org.elasticsearch.cluster.routing.ShardRouting;
import org.elasticsearch.cluster.routing.allocation.DiskThresholdSettings;
import org.elasticsearch.cluster.service.ClusterService;
import org.elasticsearch.common.Priority;
import org.elasticsearch.common.Randomness;
import org.elasticsearch.common.Strings;
import org.elasticsearch.common.UUIDs;
import org.elasticsearch.common.bytes.BytesReference;
Expand All @@ -38,6 +41,8 @@
import org.elasticsearch.core.internal.io.IOUtils;
import org.elasticsearch.env.NodeEnvironment;
import org.elasticsearch.index.Index;
import org.elasticsearch.index.engine.DocIdSeqNoAndTerm;
import org.elasticsearch.index.seqno.SeqNoStats;
import org.elasticsearch.index.seqno.SequenceNumbers;
import org.elasticsearch.index.shard.IndexShard;
import org.elasticsearch.index.shard.IndexShardTestCase;
Expand Down Expand Up @@ -76,12 +81,15 @@
import java.util.Arrays;
import java.util.Collection;
import java.util.Collections;
import java.util.HashMap;
import java.util.List;
import java.util.Locale;
import java.util.Map;
import java.util.Set;
import java.util.concurrent.CountDownLatch;
import java.util.concurrent.TimeUnit;
import java.util.function.Function;
import java.util.stream.Collectors;

import static org.elasticsearch.common.xcontent.XContentFactory.jsonBuilder;
import static org.elasticsearch.discovery.DiscoveryModule.DISCOVERY_HOSTS_PROVIDER_SETTING;
Expand All @@ -90,6 +98,7 @@
import static org.elasticsearch.test.hamcrest.ElasticsearchAssertions.assertNoFailures;
import static org.hamcrest.Matchers.empty;
import static org.hamcrest.Matchers.equalTo;
import static org.hamcrest.Matchers.greaterThanOrEqualTo;
import static org.hamcrest.Matchers.lessThanOrEqualTo;

public abstract class CcrIntegTestCase extends ESTestCase {
Expand Down Expand Up @@ -240,10 +249,12 @@ protected final ClusterHealthStatus ensureLeaderYellow(String... indices) {
}

protected final ClusterHealthStatus ensureLeaderGreen(String... indices) {
logger.info("ensure green leader indices {}", Arrays.toString(indices));
return ensureColor(clusterGroup.leaderCluster, ClusterHealthStatus.GREEN, TimeValue.timeValueSeconds(30), false, indices);
}

protected final ClusterHealthStatus ensureFollowerGreen(String... indices) {
logger.info("ensure green follower indices {}", Arrays.toString(indices));
return ensureColor(clusterGroup.followerCluster, ClusterHealthStatus.GREEN, TimeValue.timeValueSeconds(30), false, indices);
}

Expand Down Expand Up @@ -391,20 +402,54 @@ public static ResumeFollowAction.Request resumeFollow(String followerIndex) {
return request;
}

protected void assertSameDocCount(String leaderIndex, String followerIndex) throws Exception {
refresh(leaderClient(), leaderIndex);
SearchRequest request1 = new SearchRequest(leaderIndex);
request1.source(new SearchSourceBuilder().size(0));
SearchResponse response1 = leaderClient().search(request1).actionGet();
/**
* This asserts the index is fully replicated from the leader index to the follower index. It first verifies that the seq_no_stats
* on the follower equal the leader's; then verifies the existing pairs of (docId, seqNo) on the follower also equal the leader.
*/
protected void assertIndexFullyReplicatedToFollower(String leaderIndex, String followerIndex) throws Exception {
logger.info("--> asserting seq_no_stats between {} and {}", leaderIndex, followerIndex);
assertBusy(() -> {
refresh(followerClient(), followerIndex);
SearchRequest request2 = new SearchRequest(followerIndex);
request2.source(new SearchSourceBuilder().size(0));
SearchResponse response2 = followerClient().search(request2).actionGet();
assertThat(response2.getHits().getTotalHits(), equalTo(response1.getHits().getTotalHits()));
Map<Integer, SeqNoStats> leaderStats = new HashMap<>();
for (ShardStats shardStat : leaderClient().admin().indices().prepareStats(leaderIndex).clear().get().getShards()) {
if (shardStat.getSeqNoStats() == null) {
throw new AssertionError("leader seq_no_stats is not available [" + Strings.toString(shardStat) + "]");
}
leaderStats.put(shardStat.getShardRouting().shardId().id(), shardStat.getSeqNoStats());
}
Map<Integer, SeqNoStats> followerStats = new HashMap<>();
for (ShardStats shardStat : followerClient().admin().indices().prepareStats(followerIndex).clear().get().getShards()) {
if (shardStat.getSeqNoStats() == null) {
throw new AssertionError("follower seq_no_stats is not available [" + Strings.toString(shardStat) + "]");
}
followerStats.put(shardStat.getShardRouting().shardId().id(), shardStat.getSeqNoStats());
}
assertThat(leaderStats, equalTo(followerStats));
}, 60, TimeUnit.SECONDS);
logger.info("--> asserting <<docId,seqNo>> between {} and {}", leaderIndex, followerIndex);
assertBusy(() -> {
assertThat(getDocIdAndSeqNos(clusterGroup.leaderCluster, leaderIndex),
equalTo(getDocIdAndSeqNos(clusterGroup.followerCluster, followerIndex)));
}, 60, TimeUnit.SECONDS);
}

private Map<Integer, List<DocIdSeqNoAndTerm>> getDocIdAndSeqNos(InternalTestCluster cluster, String index) throws IOException {
final ClusterState state = cluster.client().admin().cluster().prepareState().get().getState();
List<ShardRouting> shardRoutings = state.routingTable().allShards(index);
Randomness.shuffle(shardRoutings);
final Map<Integer, List<DocIdSeqNoAndTerm>> docs = new HashMap<>();
for (ShardRouting shardRouting : shardRoutings) {
if (shardRouting == null || shardRouting.assignedToNode() == false || docs.containsKey(shardRouting.shardId().id())) {
continue;
}
IndexShard indexShard = cluster.getInstance(IndicesService.class, state.nodes().get(shardRouting.currentNodeId()).getName())
.indexServiceSafe(shardRouting.index()).getShard(shardRouting.id());
docs.put(shardRouting.shardId().id(), IndexShardTestCase.getDocIdAndSeqNos(indexShard).stream()
.map(d -> new DocIdSeqNoAndTerm(d.getId(), d.getSeqNo(), 1L)) // normalize primary term as the follower use its own term
.collect(Collectors.toList()));
}
return docs;
}

protected void atLeastDocsIndexed(Client client, String index, long numDocsReplicated) throws InterruptedException {
logger.info("waiting for at least [{}] documents to be indexed into index [{}]", numDocsReplicated, index);
awaitBusy(() -> {
Expand All @@ -416,6 +461,20 @@ protected void atLeastDocsIndexed(Client client, String index, long numDocsRepli
}, 60, TimeUnit.SECONDS);
}

protected void awaitGlobalCheckpointAtLeast(Client client, ShardId shardId, long minimumGlobalCheckpoint) throws Exception {
logger.info("waiting for the global checkpoint on [{}] at least [{}]", shardId, minimumGlobalCheckpoint);
assertBusy(() -> {
ShardStats stats = client.admin().indices().prepareStats(shardId.getIndexName()).clear().get()
.asMap().entrySet().stream().filter(e -> e.getKey().shardId().equals(shardId))
.map(Map.Entry::getValue).findFirst().orElse(null);
if (stats == null || stats.getSeqNoStats() == null) {
throw new AssertionError("seq_no_stats for shard [" + shardId + "] is not found"); // causes assertBusy to retry
}
assertThat(Strings.toString(stats.getSeqNoStats()),
stats.getSeqNoStats().getGlobalCheckpoint(), greaterThanOrEqualTo(minimumGlobalCheckpoint));
}, 60, TimeUnit.SECONDS);
}

protected void assertMaxSeqNoOfUpdatesIsTransferred(Index leaderIndex, Index followerIndex, int numberOfShards) throws Exception {
assertBusy(() -> {
long[] msuOnLeader = new long[numberOfShards];
Expand Down
62 changes: 40 additions & 22 deletions x-pack/plugin/ccr/src/test/java/org/elasticsearch/xpack/ccr/FollowerFailOverIT.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -19,11 +19,14 @@
import org.elasticsearch.common.unit.TimeValue;
import org.elasticsearch.common.xcontent.XContentType;
import org.elasticsearch.index.IndexSettings;
import org.elasticsearch.index.shard.ShardId;
import org.elasticsearch.test.InternalTestCluster;
import org.elasticsearch.xpack.CcrIntegTestCase;
import org.elasticsearch.xpack.core.ccr.action.PutFollowAction;

import java.util.Locale;
import java.util.concurrent.Semaphore;
import java.util.concurrent.TimeUnit;
import java.util.concurrent.atomic.AtomicBoolean;
import java.util.concurrent.atomic.AtomicInteger;

Expand All @@ -47,9 +50,17 @@ public void testFailOverOnFollower() throws Exception {
AtomicBoolean stopped = new AtomicBoolean();
Thread[] threads = new Thread[between(1, 8)];
AtomicInteger docID = new AtomicInteger();
Semaphore availableDocs = new Semaphore(0);
for (int i = 0; i < threads.length; i++) {
threads[i] = new Thread(() -> {
while (stopped.get() == false) {
try {
if (availableDocs.tryAcquire(10, TimeUnit.MILLISECONDS) == false) {
continue;
}
} catch (InterruptedException e) {
throw new AssertionError(e);
}
if (frequently()) {
String id = Integer.toString(frequently() ? docID.incrementAndGet() : between(0, 10)); // sometimes update
IndexResponse indexResponse = leaderClient().prepareIndex("leader-index", "doc", id)
Expand All @@ -64,6 +75,7 @@ public void testFailOverOnFollower() throws Exception {
});
threads[i].start();
}
availableDocs.release(between(100, 200));
PutFollowAction.Request follow = putFollow("leader-index", "follower-index");
follow.getFollowRequest().setMaxReadRequestOperationCount(randomIntBetween(32, 2048));
follow.getFollowRequest().setMaxReadRequestSize(new ByteSizeValue(randomIntBetween(1, 4096), ByteSizeUnit.KB));
Expand All @@ -74,7 +86,7 @@ public void testFailOverOnFollower() throws Exception {
logger.info("--> follow params {}", Strings.toString(follow.getFollowRequest()));
followerClient().execute(PutFollowAction.INSTANCE, follow).get();
ensureFollowerGreen("follower-index");
atLeastDocsIndexed(followerClient(), "follower-index", between(20, 60));
awaitGlobalCheckpointAtLeast(followerClient(), new ShardId(resolveFollowerIndex("follower-index"), 0), between(30, 80));
final ClusterState clusterState = getFollowerCluster().clusterService().state();
for (ShardRouting shardRouting : clusterState.routingTable().allShards("follower-index")) {
if (shardRouting.primary()) {
Expand All @@ -83,36 +95,42 @@ public void testFailOverOnFollower() throws Exception {
break;
}
}
availableDocs.release(between(50, 200));
ensureFollowerGreen("follower-index");
atLeastDocsIndexed(followerClient(), "follower-index", between(80, 150));
availableDocs.release(between(50, 200));
awaitGlobalCheckpointAtLeast(followerClient(), new ShardId(resolveFollowerIndex("follower-index"), 0), between(100, 150));
stopped.set(true);
for (Thread thread : threads) {
thread.join();
}
assertSameDocCount("leader-index", "follower-index");
assertIndexFullyReplicatedToFollower("leader-index", "follower-index");
pauseFollow("follower-index");
}

@AwaitsFix(bugUrl = "https://github.com/elastic/elasticsearch/issues/33337")
public void testFollowIndexAndCloseNode() throws Exception {
getFollowerCluster().ensureAtLeastNumDataNodes(3);
String leaderIndexSettings = getIndexSettings(3, 1, singletonMap(IndexSettings.INDEX_SOFT_DELETES_SETTING.getKey(), "true"));
assertAcked(leaderClient().admin().indices().prepareCreate("index1").setSource(leaderIndexSettings, XContentType.JSON));
ensureLeaderGreen("index1");

AtomicBoolean run = new AtomicBoolean(true);
Semaphore availableDocs = new Semaphore(0);
Thread thread = new Thread(() -> {
int counter = 0;
while (run.get()) {
final String source = String.format(Locale.ROOT, "{\"f\":%d}", counter++);
try {
leaderClient().prepareIndex("index1", "doc")
.setSource(source, XContentType.JSON)
.setTimeout(TimeValue.timeValueSeconds(1))
.get();
} catch (Exception e) {
logger.error("Error while indexing into leader index", e);
if (availableDocs.tryAcquire(10, TimeUnit.MILLISECONDS) == false) {
continue;
}
} catch (InterruptedException e) {
throw new AssertionError(e);
}
final String source = String.format(Locale.ROOT, "{\"f\":%d}", counter++);
IndexResponse indexResp = leaderClient().prepareIndex("index1", "doc")
.setSource(source, XContentType.JSON)
.setTimeout(TimeValue.timeValueSeconds(1))
.get();
logger.info("--> index id={} seq_no={}", indexResp.getId(), indexResp.getSeqNo());
}
});
thread.start();
Expand All @@ -125,19 +143,19 @@ public void testFollowIndexAndCloseNode() throws Exception {
followRequest.getFollowRequest().setMaxWriteRequestSize(new ByteSizeValue(randomIntBetween(1, 4096), ByteSizeUnit.KB));
followRequest.getFollowRequest().setMaxOutstandingWriteRequests(randomIntBetween(1, 10));
followerClient().execute(PutFollowAction.INSTANCE, followRequest).get();
logger.info("--> follow params {}", Strings.toString(followRequest.getFollowRequest()));

long maxNumDocsReplicated = Math.min(1000, randomLongBetween(followRequest.getFollowRequest().getMaxReadRequestOperationCount(),
followRequest.getFollowRequest().getMaxReadRequestOperationCount() * 10));
long minNumDocsReplicated = maxNumDocsReplicated / 3L;
logger.info("waiting for at least [{}] documents to be indexed and then stop a random data node", minNumDocsReplicated);
atLeastDocsIndexed(followerClient(), "index2", minNumDocsReplicated);
int maxOpsPerRead = followRequest.getFollowRequest().getMaxReadRequestOperationCount();
int maxNumDocsReplicated = Math.min(between(50, 500), between(maxOpsPerRead, maxOpsPerRead * 10));
availableDocs.release(maxNumDocsReplicated / 2 + 1);
atLeastDocsIndexed(followerClient(), "index2", maxNumDocsReplicated / 3);
getFollowerCluster().stopRandomNonMasterNode();
logger.info("waiting for at least [{}] documents to be indexed", maxNumDocsReplicated);
atLeastDocsIndexed(followerClient(), "index2", maxNumDocsReplicated);
availableDocs.release(maxNumDocsReplicated / 2 + 1);
atLeastDocsIndexed(followerClient(), "index2", maxNumDocsReplicated * 2 / 3);
run.set(false);
thread.join();

assertSameDocCount("index1", "index2");
assertIndexFullyReplicatedToFollower("index1", "index2");
pauseFollow("index2");
assertMaxSeqNoOfUpdatesIsTransferred(resolveLeaderIndex("index1"), resolveFollowerIndex("index2"), 3);
}
Expand Down Expand Up @@ -188,15 +206,15 @@ public void testAddNewReplicasOnFollower() throws Exception {
}
});
flushingOnFollower.start();
atLeastDocsIndexed(followerClient(), "follower-index", 50);
awaitGlobalCheckpointAtLeast(followerClient(), new ShardId(resolveFollowerIndex("follower-index"), 0), 50);
followerClient().admin().indices().prepareUpdateSettings("follower-index")
.setSettings(Settings.builder().put(IndexMetaData.SETTING_NUMBER_OF_REPLICAS, numberOfReplicas + 1).build()).get();
ensureFollowerGreen("follower-index");
atLeastDocsIndexed(followerClient(), "follower-index", 100);
awaitGlobalCheckpointAtLeast(followerClient(), new ShardId(resolveFollowerIndex("follower-index"), 0), 100);
stopped.set(true);
flushingOnFollower.join();
indexingOnLeader.join();
assertSameDocCount("leader-index", "follower-index");
assertIndexFullyReplicatedToFollower("leader-index", "follower-index");
pauseFollow("follower-index");
}

Expand Down
Loading

0 comments on commit 96af589

Please sign in to comment.