Skip to content

Commit

Permalink
Set timeout of master requests on follower to unbounded (#60070)
Browse files Browse the repository at this point in the history 
Today, a follow task will fail if the master node of the follower 
cluster is temporarily overloaded and unable to process master node
requests (such as update mapping, setting, or alias) from a follow-task
within the default timeout. This error is transient, and follow-tasks
should not abort. We can avoid this problem by setting the timeout of
master node requests on the follower cluster to unbounded.

Closes #56891
  • Loading branch information
@dnhatn
dnhatn authored Jul 27, 2020
1 parent 6398f73 commit f238d0a
Show file tree
Hide file tree
Showing 11 changed files with 108 additions and 48 deletions.
1 change: 1 addition & 0 deletions libs/core/src/main/java/org/elasticsearch/common/unit/TimeValue.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -30,6 +30,7 @@ public class TimeValue implements Comparable<TimeValue> {

public static final TimeValue MINUS_ONE = timeValueMillis(-1);
public static final TimeValue ZERO = timeValueMillis(0);
public static final TimeValue MAX_VALUE = TimeValue.timeValueNanos(Long.MAX_VALUE);

private static final long C0 = 1L;
private static final long C1 = C0 * 1000L;
Expand Down
26 changes: 15 additions & 11 deletions .../plugin/ccr/src/internalClusterTest/java/org/elasticsearch/xpack/ccr/CcrRepositoryIT.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -69,7 +69,7 @@ public void testThatRepositoryIsPutAndRemovedWhenRemoteClusterIsUpdated() throws
fail("need repository");
}

ClusterUpdateSettingsRequest putSecondCluster = new ClusterUpdateSettingsRequest();
ClusterUpdateSettingsRequest putSecondCluster = newSettingsRequest();
String address = getFollowerCluster().getDataNodeInstance(TransportService.class).boundAddress().publishAddress().toString();
putSecondCluster.persistentSettings(Settings.builder().put("cluster.remote.follower_cluster_copy.seeds", address));
assertAcked(followerClient().admin().cluster().updateSettings(putSecondCluster).actionGet());
Expand All @@ -83,19 +83,19 @@ public void testThatRepositoryIsPutAndRemovedWhenRemoteClusterIsUpdated() throws
fail("need repository");
}

ClusterUpdateSettingsRequest deleteLeaderCluster = new ClusterUpdateSettingsRequest();
ClusterUpdateSettingsRequest deleteLeaderCluster = newSettingsRequest();
deleteLeaderCluster.persistentSettings(Settings.builder().put("cluster.remote.leader_cluster.seeds", ""));
assertAcked(followerClient().admin().cluster().updateSettings(deleteLeaderCluster).actionGet());

expectThrows(RepositoryMissingException.class, () -> repositoriesService.repository(leaderClusterRepoName));

ClusterUpdateSettingsRequest deleteSecondCluster = new ClusterUpdateSettingsRequest();
ClusterUpdateSettingsRequest deleteSecondCluster = newSettingsRequest();
deleteSecondCluster.persistentSettings(Settings.builder().put("cluster.remote.follower_cluster_copy.seeds", ""));
assertAcked(followerClient().admin().cluster().updateSettings(deleteSecondCluster).actionGet());

expectThrows(RepositoryMissingException.class, () -> repositoriesService.repository(followerCopyRepoName));

ClusterUpdateSettingsRequest putLeaderRequest = new ClusterUpdateSettingsRequest();
ClusterUpdateSettingsRequest putLeaderRequest = newSettingsRequest();
address = getLeaderCluster().getDataNodeInstance(TransportService.class).boundAddress().publishAddress().toString();
putLeaderRequest.persistentSettings(Settings.builder().put("cluster.remote.leader_cluster.seeds", address));
assertAcked(followerClient().admin().cluster().updateSettings(putLeaderRequest).actionGet());
Expand All @@ -119,7 +119,7 @@ public void testThatRepositoryRecoversEmptyIndexBasedOnLeaderSettings() throws I
.put(CcrSettings.CCR_FOLLOWING_INDEX_SETTING.getKey(), true);
RestoreSnapshotRequest restoreRequest = new RestoreSnapshotRequest(leaderClusterRepoName, CcrRepository.LATEST)
.indices(leaderIndex).indicesOptions(indicesOptions).renamePattern("^(.*)$")
.renameReplacement(followerIndex).masterNodeTimeout(new TimeValue(1L, TimeUnit.HOURS))
.renameReplacement(followerIndex).masterNodeTimeout(TimeValue.MAX_VALUE)
.indexSettings(settingsBuilder);

PlainActionFuture<RestoreInfo> future = PlainActionFuture.newFuture();
Expand Down Expand Up @@ -160,7 +160,7 @@ public void testThatRepositoryRecoversEmptyIndexBasedOnLeaderSettings() throws I
}

public void testDocsAreRecovered() throws Exception {
ClusterUpdateSettingsRequest settingsRequest = new ClusterUpdateSettingsRequest();
ClusterUpdateSettingsRequest settingsRequest = newSettingsRequest();
String chunkSize = randomFrom("4KB", "128KB", "1MB");
settingsRequest.persistentSettings(Settings.builder().put(CcrSettings.RECOVERY_CHUNK_SIZE.getKey(), chunkSize));
assertAcked(followerClient().admin().cluster().updateSettings(settingsRequest).actionGet());
Expand Down Expand Up @@ -204,7 +204,7 @@ public void testDocsAreRecovered() throws Exception {
assertExpectedDocument(followerIndex, i);
}

settingsRequest = new ClusterUpdateSettingsRequest();
settingsRequest = newSettingsRequest();
ByteSizeValue defaultValue = CcrSettings.RECOVERY_CHUNK_SIZE.getDefault(Settings.EMPTY);
settingsRequest.persistentSettings(Settings.builder().put(CcrSettings.RECOVERY_CHUNK_SIZE.getKey(), defaultValue));
assertAcked(followerClient().admin().cluster().updateSettings(settingsRequest).actionGet());
Expand All @@ -213,7 +213,7 @@ public void testDocsAreRecovered() throws Exception {
public void testRateLimitingIsEmployed() throws Exception {
boolean followerRateLimiting = randomBoolean();

ClusterUpdateSettingsRequest settingsRequest = new ClusterUpdateSettingsRequest();
ClusterUpdateSettingsRequest settingsRequest = newSettingsRequest();
settingsRequest.persistentSettings(Settings.builder().put(CcrSettings.RECOVERY_MAX_BYTES_PER_SECOND.getKey(), "10K"));
if (followerRateLimiting) {
assertAcked(followerClient().admin().cluster().updateSettings(settingsRequest).actionGet());
Expand Down Expand Up @@ -257,7 +257,7 @@ public void testRateLimitingIsEmployed() throws Exception {
.put(CcrSettings.CCR_FOLLOWING_INDEX_SETTING.getKey(), true);
RestoreSnapshotRequest restoreRequest = new RestoreSnapshotRequest(leaderClusterRepoName, CcrRepository.LATEST)
.indices(leaderIndex).indicesOptions(indicesOptions).renamePattern("^(.*)$")
.renameReplacement(followerIndex).masterNodeTimeout(new TimeValue(1L, TimeUnit.HOURS))
.renameReplacement(followerIndex).masterNodeTimeout(TimeValue.MAX_VALUE)
.indexSettings(settingsBuilder);

PlainActionFuture<RestoreInfo> future = PlainActionFuture.newFuture();
Expand All @@ -270,7 +270,7 @@ public void testRateLimitingIsEmployed() throws Exception {
assertTrue(restoreSources.stream().anyMatch(cr -> cr.getThrottleTime() > 0));
}

settingsRequest = new ClusterUpdateSettingsRequest();
settingsRequest = newSettingsRequest();
ByteSizeValue defaultValue = CcrSettings.RECOVERY_MAX_BYTES_PER_SECOND.getDefault(Settings.EMPTY);
settingsRequest.persistentSettings(Settings.builder().put(CcrSettings.RECOVERY_MAX_BYTES_PER_SECOND.getKey(), defaultValue));
if (followerRateLimiting) {
Expand All @@ -281,7 +281,7 @@ public void testRateLimitingIsEmployed() throws Exception {
}

public void testIndividualActionsTimeout() throws Exception {
ClusterUpdateSettingsRequest settingsRequest = new ClusterUpdateSettingsRequest();
ClusterUpdateSettingsRequest settingsRequest = newSettingsRequest();
TimeValue timeValue = TimeValue.timeValueMillis(100);
settingsRequest.persistentSettings(Settings.builder().put(CcrSettings.INDICES_RECOVERY_ACTION_TIMEOUT_SETTING.getKey(), timeValue));
assertAcked(followerClient().admin().cluster().updateSettings(settingsRequest).actionGet());
Expand Down Expand Up @@ -359,6 +359,10 @@ public void testIndividualActionsTimeout() throws Exception {
}
}

private ClusterUpdateSettingsRequest newSettingsRequest() {
return new ClusterUpdateSettingsRequest().masterNodeTimeout(TimeValue.MAX_VALUE);
}

public void testFollowerMappingIsUpdated() throws IOException {
String leaderClusterRepoName = CcrRepository.NAME_PREFIX + "leader_cluster";
String leaderIndex = "index1";
Expand Down
11 changes: 6 additions & 5 deletions ...gin/ccr/src/internalClusterTest/java/org/elasticsearch/xpack/ccr/CcrRetentionLeaseIT.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -119,7 +119,7 @@ private RestoreSnapshotRequest setUpRestoreSnapshotRequest(
final int numberOfReplicas,
final String followerIndex,
final int numberOfDocuments) throws IOException {
final ClusterUpdateSettingsRequest settingsRequest = new ClusterUpdateSettingsRequest();
final ClusterUpdateSettingsRequest settingsRequest = new ClusterUpdateSettingsRequest().masterNodeTimeout(TimeValue.MAX_VALUE);
final String chunkSize = new ByteSizeValue(randomFrom(4, 128, 1024), ByteSizeUnit.KB).getStringRep();
settingsRequest.persistentSettings(Settings.builder().put(CcrSettings.RECOVERY_CHUNK_SIZE.getKey(), chunkSize));
assertAcked(followerClient().admin().cluster().updateSettings(settingsRequest).actionGet());
Expand All @@ -129,7 +129,8 @@ private RestoreSnapshotRequest setUpRestoreSnapshotRequest(
final Map<String, String> additionalSettings = new HashMap<>();
additionalSettings.put(IndexService.RETENTION_LEASE_SYNC_INTERVAL_SETTING.getKey(), TimeValue.timeValueMillis(200).getStringRep());
final String leaderIndexSettings = getIndexSettings(numberOfShards, numberOfReplicas, additionalSettings);
assertAcked(leaderClient().admin().indices().prepareCreate(leaderIndex).setSource(leaderIndexSettings, XContentType.JSON));
assertAcked(leaderClient().admin().indices().prepareCreate(leaderIndex)
.setMasterNodeTimeout(TimeValue.MAX_VALUE).setSource(leaderIndexSettings, XContentType.JSON));
ensureLeaderGreen(leaderIndex);

logger.info("indexing [{}] docs", numberOfDocuments);
Expand All @@ -152,7 +153,7 @@ private RestoreSnapshotRequest setUpRestoreSnapshotRequest(
.indicesOptions(indicesOptions)
.renamePattern("^(.*)$")
.renameReplacement(followerIndex)
.masterNodeTimeout(new TimeValue(1L, TimeUnit.HOURS));
.masterNodeTimeout(TimeValue.MAX_VALUE);
}

public void testRetentionLeaseIsTakenAtTheStartOfRecovery() throws Exception {
Expand Down Expand Up @@ -470,7 +471,7 @@ public void testUnfollowFailsToRemoveRetentionLeases() throws Exception {
ensureFollowerGreen(true, followerIndex);

pauseFollow(followerIndex);
followerClient().admin().indices().close(new CloseIndexRequest(followerIndex)).actionGet();
followerClient().admin().indices().close(new CloseIndexRequest(followerIndex).masterNodeTimeout(TimeValue.MAX_VALUE)).actionGet();

// we will disrupt requests to remove retention leases for these random shards
final Set<Integer> shardIds =
Expand Down Expand Up @@ -944,7 +945,7 @@ public void testForgetFollower() throws Exception {
ensureFollowerGreen(true, followerIndex);

pauseFollow(followerIndex);
followerClient().admin().indices().close(new CloseIndexRequest(followerIndex)).actionGet();
followerClient().admin().indices().close(new CloseIndexRequest(followerIndex).masterNodeTimeout(TimeValue.MAX_VALUE)).actionGet();

final ClusterStateResponse followerIndexClusterState =
followerClient().admin().cluster().prepareState().clear().setMetadata(true).setIndices(followerIndex).get();
Expand Down
4 changes: 2 additions & 2 deletions ...in/ccr/src/internalClusterTest/java/org/elasticsearch/xpack/ccr/CloseFollowerIndexIT.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -96,7 +96,7 @@ public void testCloseAndReopenFollowerIndex() throws Exception {

atLeastDocsIndexed(followerClient(), "index2", 32);

CloseIndexRequest closeIndexRequest = new CloseIndexRequest("index2");
CloseIndexRequest closeIndexRequest = new CloseIndexRequest("index2").masterNodeTimeout(TimeValue.MAX_VALUE);
closeIndexRequest.waitForActiveShards(ActiveShardCount.NONE);
AcknowledgedResponse response = followerClient().admin().indices().close(closeIndexRequest).get();
assertThat(response.isAcknowledged(), is(true));
Expand All @@ -111,7 +111,7 @@ public void testCloseAndReopenFollowerIndex() throws Exception {
thread.join();
}

assertAcked(followerClient().admin().indices().open(new OpenIndexRequest("index2")).get());
assertAcked(followerClient().admin().indices().open(new OpenIndexRequest("index2").masterNodeTimeout(TimeValue.MAX_VALUE)).get());

clusterState = followerClient().admin().cluster().prepareState().get().getState();
assertThat(clusterState.metadata().index("index2").getState(), is(IndexMetadata.State.OPEN));
Expand Down
2 changes: 1 addition & 1 deletion ...ugin/ccr/src/internalClusterTest/java/org/elasticsearch/xpack/ccr/FollowerFailOverIT.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -223,7 +223,7 @@ public void testAddNewReplicasOnFollower() throws Exception {
});
flushingOnFollower.start();
awaitGlobalCheckpointAtLeast(followerClient(), new ShardId(resolveFollowerIndex("follower-index"), 0), 50);
followerClient().admin().indices().prepareUpdateSettings("follower-index")
followerClient().admin().indices().prepareUpdateSettings("follower-index").setMasterNodeTimeout(TimeValue.MAX_VALUE)
.setSettings(Settings.builder().put(IndexMetadata.SETTING_NUMBER_OF_REPLICAS, numberOfReplicas + 1).build()).get();
ensureFollowerGreen("follower-index");
awaitGlobalCheckpointAtLeast(followerClient(), new ShardId(resolveFollowerIndex("follower-index"), 0), 100);
Expand Down
Loading

0 comments on commit f238d0a

Please sign in to comment.