Skip to content

[rpc] Add client-side RPC request deadlines #4348

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
hutiefang76 wants to merge 1 commit into
base: main
Choose a base branch
from
Open

[rpc] Add client-side RPC request deadlines #4348

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 4 additions & 0 deletions fluss-rpc/src/main/java/org/apache/fluss/rpc/netty/client/NettyClient.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -76,6 +76,8 @@ public final class NettyClient implements RpcClient {

private final Supplier<ClientAuthenticator> authenticatorSupplier;

private final long requestTimeoutMs;

private volatile boolean isClosed = false;

public NettyClient(Configuration conf, ClientMetricGroup clientMetricGroup) {
Expand All @@ -87,6 +89,7 @@ public NettyClient(Configuration conf, ClientMetricGroup clientMetricGroup) {
conf.getInt(ConfigOptions.NETTY_CLIENT_NUM_NETWORK_THREADS),
"fluss-netty-client");
int connectTimeoutMs = (int) conf.get(ConfigOptions.CLIENT_CONNECT_TIMEOUT).toMillis();
this.requestTimeoutMs = conf.get(ConfigOptions.CLIENT_REQUEST_TIMEOUT).toMillis();
int connectionMaxIdle =
(int) conf.get(ConfigOptions.NETTY_CONNECTION_MAX_IDLE_TIME).getSeconds();
boolean preferHeap =
Expand Down Expand Up @@ -191,6 +194,7 @@ private ServerConnection getOrCreateConnection(ServerNode node) {
node,
clientMetricGroup,
authenticatorSupplier.get(),
requestTimeoutMs,
(con, ignore) -> connections.remove(serverId, con));
});
}
Expand Down
44 changes: 33 additions & 11 deletions fluss-rpc/src/main/java/org/apache/fluss/rpc/netty/client/ServerConnection.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -48,6 +48,7 @@
import org.apache.fluss.shaded.netty4.io.netty.channel.ChannelFuture;
import org.apache.fluss.shaded.netty4.io.netty.channel.ChannelFutureListener;
import org.apache.fluss.utils.ExponentialBackoff;
import org.apache.fluss.utils.concurrent.FutureUtils;

import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
Expand Down Expand Up @@ -85,6 +86,7 @@ final class ServerConnection {
private final CompletableFuture<Void> closeFuture = new CompletableFuture<>();
private final ConnectionMetrics connectionMetrics;
private final ClientAuthenticator authenticator;
private final long requestTimeoutMs;
private final ExponentialBackoff backoff;
private final Object lock = new Object();

Expand Down Expand Up @@ -114,11 +116,13 @@ final class ServerConnection {
ServerNode node,
ClientMetricGroup clientMetricGroup,
ClientAuthenticator authenticator,
long requestTimeoutMs,
BiConsumer<ServerConnection, Throwable> closeCallback) {
this.node = node;
this.state = ConnectionState.CONNECTING;
this.connectionMetrics = clientMetricGroup.createConnectionMetricGroup(node.uid());
this.authenticator = authenticator;
this.requestTimeoutMs = requestTimeoutMs;
this.backoff = new ExponentialBackoff(100L, 2, 5000L, 0.2);
whenClose(closeCallback);

Expand Down Expand Up @@ -360,23 +364,41 @@ private CompletableFuture<ApiMessage> doSend(
return responseFuture;
}

FutureUtils.orTimeout(
inflight.responseFuture,
requestTimeoutMs,
TimeUnit.MILLISECONDS,
String.format(
"Timed out waiting for response from node %s after %d ms.",
node, requestTimeoutMs));
inflight.responseFuture.whenComplete(
(ignored, throwable) -> {
if (inflightRequests.remove(inflight.requestId, inflight)) {
connectionMetrics.updateMetricsAfterGetResponse(
apiKey, inflight.requestStartTime, 0);
}
});

channel.writeAndFlush(byteBuf)
.addListener(
(ChannelFutureListener)
future -> {
if (!future.isSuccess()) {
connectionMetrics.updateMetricsAfterGetResponse(
apiKey, inflight.requestStartTime, 0);
Throwable cause = future.cause();
if (cause instanceof IOException) {
// when server close the channel, the cause will be
// IOException, if the cause is IOException, we wrap
// it as retryable NetworkException to retry to
// connect
cause = new NetworkException(cause);
if (inflightRequests.remove(
inflight.requestId, inflight)) {
connectionMetrics.updateMetricsAfterGetResponse(
apiKey, inflight.requestStartTime, 0);
Throwable cause = future.cause();
if (cause instanceof IOException) {
// when server close the channel, the cause will
// be IOException, if the cause is IOException,
// we wrap it as retryable NetworkException to
// retry to connect
cause = new NetworkException(cause);
}
inflight.responseFuture.completeExceptionally(
cause);
}
inflight.responseFuture.completeExceptionally(cause);
inflightRequests.remove(inflight.requestId);
}
});
return inflight.responseFuture;
Expand Down
50 changes: 49 additions & 1 deletion fluss-rpc/src/test/java/org/apache/fluss/rpc/netty/client/NettyClientTest.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -27,6 +27,7 @@
import org.apache.fluss.rpc.TestingGatewayService;
import org.apache.fluss.rpc.messages.ApiMessage;
import org.apache.fluss.rpc.messages.ApiVersionsRequest;
import org.apache.fluss.rpc.messages.ApiVersionsResponse;
import org.apache.fluss.rpc.messages.GetTableInfoRequest;
import org.apache.fluss.rpc.messages.LookupRequest;
import org.apache.fluss.rpc.messages.PbLookupReqForBucket;
Expand All @@ -43,6 +44,7 @@
import org.junit.jupiter.api.Test;

import java.net.ConnectException;
import java.time.Duration;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.Collections;
Expand All @@ -51,6 +53,9 @@
import java.util.Set;
import java.util.concurrent.CompletableFuture;
import java.util.concurrent.ExecutionException;
import java.util.concurrent.TimeUnit;
import java.util.concurrent.TimeoutException;
import java.util.concurrent.atomic.AtomicInteger;

import static org.apache.fluss.utils.NetUtils.getAvailablePort;
import static org.assertj.core.api.Assertions.assertThat;
Expand Down Expand Up @@ -99,6 +104,44 @@ void testSendIncompleteRequest() {
.hasRootCauseMessage("Some required fields are missing");
}

@Test
void testClientRequestTimeout() throws Exception {
nettyClient.close();
nettyServer.close();

conf.set(ConfigOptions.CLIENT_REQUEST_TIMEOUT, Duration.ofMillis(100));
conf.set(ConfigOptions.NETTY_CONNECTION_MAX_IDLE_TIME, Duration.ofSeconds(2));
nettyClient = new NettyClient(conf, TestingClientMetricGroup.newInstance());

AtomicInteger apiVersionsRequests = new AtomicInteger();
buildNettyServer(
1,
new TestingGatewayService() {
@Override
public CompletableFuture<ApiVersionsResponse> apiVersions(
ApiVersionsRequest request) {
if (apiVersionsRequests.incrementAndGet() == 1) {
return super.apiVersions(request);
}
return new CompletableFuture<>();
}
});

ApiVersionsRequest request =
new ApiVersionsRequest()
.setClientSoftwareName("testing_client")
.setClientSoftwareVersion("1.0");
CompletableFuture<ApiMessage> response =
nettyClient.sendRequest(serverNode, ApiKeys.API_VERSIONS, request);

assertThatThrownBy(() -> response.get(1, TimeUnit.SECONDS))
.isInstanceOf(ExecutionException.class)
.hasRootCauseInstanceOf(TimeoutException.class)
.rootCause()
.hasMessageContaining("Timed out waiting for response from node");
assertThat(nettyClient.connections().get(serverNode.uid()).numInflightRequests()).isZero();
}

@Test
void testSendRequestToWrongServerType() {
LookupRequest lookupRequest = new LookupRequest().setTableId(1);
Expand Down Expand Up @@ -254,11 +297,16 @@ void testExceptionWhenInitializeServerConnection() throws Exception {
}

private void buildNettyServer(int serverId) throws Exception {
buildNettyServer(serverId, new TestingGatewayService());
}

private void buildNettyServer(int serverId, TestingGatewayService gatewayService)
throws Exception {
try (NetUtils.Port availablePort = getAvailablePort()) {
serverNode =
new ServerNode(
serverId, "localhost", availablePort.getPort(), ServerType.COORDINATOR);
service = new TestingGatewayService();
service = gatewayService;
MetricGroup metricGroup = NOPMetricsGroup.newInstance();
nettyServer =
new NettyServer(
Expand Down
23 changes: 21 additions & 2 deletions fluss-rpc/src/test/java/org/apache/fluss/rpc/netty/client/ServerConnectionTest.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,6 +20,7 @@
import org.apache.fluss.cluster.Endpoint;
import org.apache.fluss.cluster.ServerNode;
import org.apache.fluss.cluster.ServerType;
import org.apache.fluss.config.ConfigOptions;
import org.apache.fluss.config.Configuration;
import org.apache.fluss.exception.DisconnectException;
import org.apache.fluss.exception.InvalidServerTypeException;
Expand Down Expand Up @@ -96,6 +97,9 @@
/** Test for {@link ServerConnection}. */
public class ServerConnectionTest {

private static final long REQUEST_TIMEOUT_MS =
ConfigOptions.CLIENT_REQUEST_TIMEOUT.defaultValue().toMillis();

private EventLoopGroup eventLoopGroup;
private Bootstrap bootstrap;
private ClientAuthenticator clientAuthenticator;
Expand Down Expand Up @@ -139,6 +143,7 @@ void testConnectionClose() {
serverNode,
TestingClientMetricGroup.newInstance(),
clientAuthenticator,
REQUEST_TIMEOUT_MS,
(con, ignore) -> {});
ConnectionState connectionState = connection.getConnectionState();
assertThat(connectionState).isEqualTo(ConnectionState.CONNECTING);
Expand Down Expand Up @@ -169,10 +174,20 @@ void testConnectionMetrics() throws ExecutionException, InterruptedException {
ClientMetricGroup client = new ClientMetricGroup(metricRegistry, "client");
ServerConnection connection =
new ServerConnection(
bootstrap, serverNode, client, clientAuthenticator, (con, ignore) -> {});
bootstrap,
serverNode,
client,
clientAuthenticator,
REQUEST_TIMEOUT_MS,
(con, ignore) -> {});
ServerConnection connection2 =
new ServerConnection(
bootstrap, serverNode2, client, clientAuthenticator, (con, ignore) -> {});
bootstrap,
serverNode2,
client,
clientAuthenticator,
REQUEST_TIMEOUT_MS,
(con, ignore) -> {});
LookupRequest request = new LookupRequest().setTableId(1);
PbLookupReqForBucket pbLookupReqForBucket = request.addBucketsReq();
pbLookupReqForBucket.setBucketId(1);
Expand Down Expand Up @@ -222,6 +237,7 @@ public ChannelFuture connect(String host, int port) {
serverNode,
TestingClientMetricGroup.newInstance(),
clientAuthenticator,
REQUEST_TIMEOUT_MS,
(con, ignore) -> {});
try {
OutOfMemoryError error = new OutOfMemoryError("Direct buffer memory");
Expand Down Expand Up @@ -272,6 +288,7 @@ public ChannelFuture connect(String host, int port) {
wrongServerTypeNode,
TestingClientMetricGroup.newInstance(),
clientAuthenticator,
REQUEST_TIMEOUT_MS,
(con, ignore) -> {});

// Pending request will be rejected with InvalidServerTypeException which is
Expand Down Expand Up @@ -304,6 +321,7 @@ void testRejectBucketCountChangeForOldServer() throws Exception {
serverNode,
TestingClientMetricGroup.newInstance(),
clientAuthenticator,
REQUEST_TIMEOUT_MS,
(con, ignore) -> {});
try {
connection.validateVersionCompatibility(
Expand Down Expand Up @@ -334,6 +352,7 @@ void testRejectHistoricalWritesForOldServer() throws Exception {
serverNode,
TestingClientMetricGroup.newInstance(),
clientAuthenticator,
REQUEST_TIMEOUT_MS,
(con, ignore) -> {});
try {
assertThat(connection.send(ApiKeys.PUT_KV, putKvRequest(null)).get())
Expand Down