scwhittle commented on code in PR #34053: URL: https://github.com/apache/beam/pull/34053#discussion_r2011744694
########## runners/google-cloud-dataflow-java/worker/src/main/java/org/apache/beam/runners/dataflow/worker/streaming/harness/BackendWorkerMetadataVendor.java: ########## @@ -0,0 +1,179 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF 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.apache.beam.runners.dataflow.worker.streaming.harness; + +import java.util.concurrent.CompletableFuture; +import java.util.concurrent.CountDownLatch; +import java.util.concurrent.ExecutorService; +import java.util.concurrent.Executors; +import java.util.concurrent.TimeUnit; +import java.util.concurrent.atomic.AtomicBoolean; +import java.util.function.Consumer; +import javax.annotation.concurrent.ThreadSafe; +import org.apache.beam.runners.dataflow.worker.windmill.Windmill.WorkerMetadataResponse; +import org.apache.beam.runners.dataflow.worker.windmill.WindmillEndpoints; +import org.apache.beam.runners.dataflow.worker.windmill.client.WindmillStream.GetWorkerMetadataStream; +import org.apache.beam.vendor.guava.v32_1_2_jre.com.google.common.util.concurrent.MoreExecutors; +import org.apache.beam.vendor.guava.v32_1_2_jre.com.google.common.util.concurrent.ThreadFactoryBuilder; +import org.slf4j.Logger; +import org.slf4j.LoggerFactory; + +/** + * Manages vending the backend worker metadata to a single consumer. + * + * <p>Internally restarts the underlying {@link GetWorkerMetadataStream} to gracefully close the + * stream and prevent gRPC {@link + * org.apache.beam.vendor.grpc.v1p69p0.io.grpc.Status#DEADLINE_EXCEEDED} statuses. + */ +@ThreadSafe +final class BackendWorkerMetadataVendor { + private static final Logger LOG = LoggerFactory.getLogger(BackendWorkerMetadataVendor.class); + private static final int GET_WORKER_METADATA_STREAM_TIMEOUT_MINUTES = 3; + private static final String BACKEND_WORKER_METADATA_VENDOR_THREAD_NAME = + "BackendWorkerMetadataVendorThread"; + private static final WorkerMetadataResponse NO_WORKER_METADATA = + WorkerMetadataResponse.getDefaultInstance(); + + private final WorkerMetadataStreamFactory getWorkerMetadataStreamFactory; + private final ExecutorService workerMetadataFetcher; + private final AtomicBoolean isRunning = new AtomicBoolean(false); + + private BackendWorkerMetadataVendor( + WorkerMetadataStreamFactory getWorkerMetadataStreamFactory, + ExecutorService workerMetadataFetcher) { + this.getWorkerMetadataStreamFactory = getWorkerMetadataStreamFactory; + this.workerMetadataFetcher = workerMetadataFetcher; + } + + static BackendWorkerMetadataVendor create( + WorkerMetadataStreamFactory getWorkerMetadataStreamFactory) { + return new BackendWorkerMetadataVendor( + getWorkerMetadataStreamFactory, + Executors.newSingleThreadExecutor( + new ThreadFactoryBuilder() + .setNameFormat(BACKEND_WORKER_METADATA_VENDOR_THREAD_NAME) + .build())); + } + + /** + * Starts the vending of backend worker metadata, passing it to endpointsConsumer for processing. + * Returns a {@link CompletableFuture} that can be used to block until the initial {@link + * GetWorkerMetadataStream} has been started. + */ + CompletableFuture<Void> start(Consumer<WindmillEndpoints> endpointsConsumer) { + if (isRunning.compareAndSet(false, true)) { Review Comment: can we make the states of this class simpler? How about just allow start to be called once, and stop to be called once after start? The atomic bool makes it seem like we support start/stop being called concurrently and repeatedly but it isn't really supported since if we actually had concurrent calls they could race since the bool doesn't guard everything else in the function. ########## runners/google-cloud-dataflow-java/worker/src/main/java/org/apache/beam/runners/dataflow/worker/windmill/client/grpc/GrpcDispatcherClient.java: ########## @@ -132,25 +132,23 @@ ImmutableSet<HostAndPort> getDispatcherEndpoints() { /** Will block the calling thread until the initial endpoints are present. */ public CloudWindmillMetadataServiceV1Alpha1Stub getWindmillMetadataServiceStubBlocking() { - boolean initialized = false; long secondsWaited = 0; - while (!initialized) { - LOG.info( - "Blocking until Windmill Service endpoint has been set. " - + "Currently waited for [{}] seconds.", - secondsWaited); - try { - initialized = onInitializedEndpoints.await(10, TimeUnit.SECONDS); + try { + while (!onInitializedEndpoints.await(10, TimeUnit.SECONDS)) { secondsWaited += 10; - } catch (InterruptedException e) { - LOG.error( - "Interrupted while waiting for initial Windmill Service endpoints. " - + "These endpoints are required to do any pipeline processing.", - e); + LOG.debug( + "Blocking until Windmill Service endpoint has been set. " + + "Currently waited for [{}] seconds.", + secondsWaited); } + } catch (InterruptedException e) { + LOG.error( + "Interrupted while waiting for initial Windmill Service endpoints. " + + "These endpoints are required to do any pipeline processing.", + e); Review Comment: throw a runtime exception? if you just fall through below it seems lik eit is just going to fail for other reasons ########## runners/google-cloud-dataflow-java/worker/src/main/java/org/apache/beam/runners/dataflow/worker/streaming/harness/BackendWorkerMetadataVendor.java: ########## @@ -0,0 +1,179 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF 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.apache.beam.runners.dataflow.worker.streaming.harness; + +import java.util.concurrent.CompletableFuture; +import java.util.concurrent.CountDownLatch; +import java.util.concurrent.ExecutorService; +import java.util.concurrent.Executors; +import java.util.concurrent.TimeUnit; +import java.util.concurrent.atomic.AtomicBoolean; +import java.util.function.Consumer; +import javax.annotation.concurrent.ThreadSafe; +import org.apache.beam.runners.dataflow.worker.windmill.Windmill.WorkerMetadataResponse; +import org.apache.beam.runners.dataflow.worker.windmill.WindmillEndpoints; +import org.apache.beam.runners.dataflow.worker.windmill.client.WindmillStream.GetWorkerMetadataStream; +import org.apache.beam.vendor.guava.v32_1_2_jre.com.google.common.util.concurrent.MoreExecutors; +import org.apache.beam.vendor.guava.v32_1_2_jre.com.google.common.util.concurrent.ThreadFactoryBuilder; +import org.slf4j.Logger; +import org.slf4j.LoggerFactory; + +/** + * Manages vending the backend worker metadata to a single consumer. + * + * <p>Internally restarts the underlying {@link GetWorkerMetadataStream} to gracefully close the + * stream and prevent gRPC {@link + * org.apache.beam.vendor.grpc.v1p69p0.io.grpc.Status#DEADLINE_EXCEEDED} statuses. + */ +@ThreadSafe +final class BackendWorkerMetadataVendor { + private static final Logger LOG = LoggerFactory.getLogger(BackendWorkerMetadataVendor.class); + private static final int GET_WORKER_METADATA_STREAM_TIMEOUT_MINUTES = 3; + private static final String BACKEND_WORKER_METADATA_VENDOR_THREAD_NAME = + "BackendWorkerMetadataVendorThread"; + private static final WorkerMetadataResponse NO_WORKER_METADATA = + WorkerMetadataResponse.getDefaultInstance(); + + private final WorkerMetadataStreamFactory getWorkerMetadataStreamFactory; + private final ExecutorService workerMetadataFetcher; Review Comment: nit: can we name this executor or thread instead of fetcher? I think it will help readability ########## runners/google-cloud-dataflow-java/worker/src/main/java/org/apache/beam/runners/dataflow/worker/streaming/harness/BackendWorkerMetadataVendor.java: ########## @@ -0,0 +1,179 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF 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.apache.beam.runners.dataflow.worker.streaming.harness; + +import java.util.concurrent.CompletableFuture; +import java.util.concurrent.CountDownLatch; +import java.util.concurrent.ExecutorService; +import java.util.concurrent.Executors; +import java.util.concurrent.TimeUnit; +import java.util.concurrent.atomic.AtomicBoolean; +import java.util.function.Consumer; +import javax.annotation.concurrent.ThreadSafe; +import org.apache.beam.runners.dataflow.worker.windmill.Windmill.WorkerMetadataResponse; +import org.apache.beam.runners.dataflow.worker.windmill.WindmillEndpoints; +import org.apache.beam.runners.dataflow.worker.windmill.client.WindmillStream.GetWorkerMetadataStream; +import org.apache.beam.vendor.guava.v32_1_2_jre.com.google.common.util.concurrent.MoreExecutors; +import org.apache.beam.vendor.guava.v32_1_2_jre.com.google.common.util.concurrent.ThreadFactoryBuilder; +import org.slf4j.Logger; +import org.slf4j.LoggerFactory; + +/** + * Manages vending the backend worker metadata to a single consumer. + * + * <p>Internally restarts the underlying {@link GetWorkerMetadataStream} to gracefully close the + * stream and prevent gRPC {@link + * org.apache.beam.vendor.grpc.v1p69p0.io.grpc.Status#DEADLINE_EXCEEDED} statuses. + */ +@ThreadSafe +final class BackendWorkerMetadataVendor { + private static final Logger LOG = LoggerFactory.getLogger(BackendWorkerMetadataVendor.class); + private static final int GET_WORKER_METADATA_STREAM_TIMEOUT_MINUTES = 3; + private static final String BACKEND_WORKER_METADATA_VENDOR_THREAD_NAME = + "BackendWorkerMetadataVendorThread"; + private static final WorkerMetadataResponse NO_WORKER_METADATA = + WorkerMetadataResponse.getDefaultInstance(); + + private final WorkerMetadataStreamFactory getWorkerMetadataStreamFactory; + private final ExecutorService workerMetadataFetcher; + private final AtomicBoolean isRunning = new AtomicBoolean(false); + + private BackendWorkerMetadataVendor( + WorkerMetadataStreamFactory getWorkerMetadataStreamFactory, + ExecutorService workerMetadataFetcher) { + this.getWorkerMetadataStreamFactory = getWorkerMetadataStreamFactory; + this.workerMetadataFetcher = workerMetadataFetcher; + } + + static BackendWorkerMetadataVendor create( + WorkerMetadataStreamFactory getWorkerMetadataStreamFactory) { + return new BackendWorkerMetadataVendor( + getWorkerMetadataStreamFactory, + Executors.newSingleThreadExecutor( + new ThreadFactoryBuilder() + .setNameFormat(BACKEND_WORKER_METADATA_VENDOR_THREAD_NAME) + .build())); + } + + /** + * Starts the vending of backend worker metadata, passing it to endpointsConsumer for processing. + * Returns a {@link CompletableFuture} that can be used to block until the initial {@link + * GetWorkerMetadataStream} has been started. + */ + CompletableFuture<Void> start(Consumer<WindmillEndpoints> endpointsConsumer) { + if (isRunning.compareAndSet(false, true)) { + LOG.debug("Starting WorkerMetadataVendor..."); + + CountDownLatch initialStreamStarted = new CountDownLatch(1); + + workerMetadataFetcher.execute( + () -> { + boolean isInitialWorkerMetadata = true; + WorkerMetadataResponse initialWorkerMetadata = NO_WORKER_METADATA; + + while (isRunning.get()) { + GetWorkerMetadataStream getWorkerMetadataStream = + getWorkerMetadataStreamFactory.create(initialWorkerMetadata, endpointsConsumer); + LOG.debug( + "Starting GetWorkerMetadataStream w/ metadata version {}.", + initialWorkerMetadata.getMetadataVersion()); + getWorkerMetadataStream.start(); + + if (isInitialWorkerMetadata) { + isInitialWorkerMetadata = false; + initialStreamStarted.countDown(); + } + + // Await stream termination and propagate the most current worker metadata to start + // the next stream. + initialWorkerMetadata = awaitGracefulTermination(getWorkerMetadataStream); + LOG.debug( + "Current GetWorkerMetadataStream terminated. Propagating metadata version {} to the next stream.", + initialWorkerMetadata.getMetadataVersion()); + } + }); + + return CompletableFuture.runAsync( + () -> { + int secondsWaited = 0; + try { + while (!initialStreamStarted.await(10, TimeUnit.SECONDS) && isRunning.get()) { + secondsWaited += 10; + LOG.debug( + "Waited {} seconds for initial GetWorkerMetadataStream to start.", + secondsWaited); + } + } catch (InterruptedException e) { + LOG.debug("Interrupted waiting for initial GetWorkerMetadataStream to start."); + Thread.currentThread().interrupt(); + } + }, + // Run this on the calling thread. Review Comment: I think this isn't going to do what you want, I think that it is going to run lambda inside of runAsync here beneath start, not when it is joined. Instead of returning a future from start, what about just making a separate awaitStreamStarted method that can be called after start by whatever thread wants to block? ########## runners/google-cloud-dataflow-java/worker/src/main/java/org/apache/beam/runners/dataflow/worker/streaming/harness/BackendWorkerMetadataVendor.java: ########## @@ -0,0 +1,179 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF 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.apache.beam.runners.dataflow.worker.streaming.harness; + +import java.util.concurrent.CompletableFuture; +import java.util.concurrent.CountDownLatch; +import java.util.concurrent.ExecutorService; +import java.util.concurrent.Executors; +import java.util.concurrent.TimeUnit; +import java.util.concurrent.atomic.AtomicBoolean; +import java.util.function.Consumer; +import javax.annotation.concurrent.ThreadSafe; +import org.apache.beam.runners.dataflow.worker.windmill.Windmill.WorkerMetadataResponse; +import org.apache.beam.runners.dataflow.worker.windmill.WindmillEndpoints; +import org.apache.beam.runners.dataflow.worker.windmill.client.WindmillStream.GetWorkerMetadataStream; +import org.apache.beam.vendor.guava.v32_1_2_jre.com.google.common.util.concurrent.MoreExecutors; +import org.apache.beam.vendor.guava.v32_1_2_jre.com.google.common.util.concurrent.ThreadFactoryBuilder; +import org.slf4j.Logger; +import org.slf4j.LoggerFactory; + +/** + * Manages vending the backend worker metadata to a single consumer. + * + * <p>Internally restarts the underlying {@link GetWorkerMetadataStream} to gracefully close the + * stream and prevent gRPC {@link + * org.apache.beam.vendor.grpc.v1p69p0.io.grpc.Status#DEADLINE_EXCEEDED} statuses. + */ +@ThreadSafe +final class BackendWorkerMetadataVendor { + private static final Logger LOG = LoggerFactory.getLogger(BackendWorkerMetadataVendor.class); + private static final int GET_WORKER_METADATA_STREAM_TIMEOUT_MINUTES = 3; + private static final String BACKEND_WORKER_METADATA_VENDOR_THREAD_NAME = + "BackendWorkerMetadataVendorThread"; + private static final WorkerMetadataResponse NO_WORKER_METADATA = + WorkerMetadataResponse.getDefaultInstance(); + + private final WorkerMetadataStreamFactory getWorkerMetadataStreamFactory; + private final ExecutorService workerMetadataFetcher; + private final AtomicBoolean isRunning = new AtomicBoolean(false); + + private BackendWorkerMetadataVendor( + WorkerMetadataStreamFactory getWorkerMetadataStreamFactory, + ExecutorService workerMetadataFetcher) { + this.getWorkerMetadataStreamFactory = getWorkerMetadataStreamFactory; + this.workerMetadataFetcher = workerMetadataFetcher; + } + + static BackendWorkerMetadataVendor create( + WorkerMetadataStreamFactory getWorkerMetadataStreamFactory) { + return new BackendWorkerMetadataVendor( + getWorkerMetadataStreamFactory, + Executors.newSingleThreadExecutor( + new ThreadFactoryBuilder() + .setNameFormat(BACKEND_WORKER_METADATA_VENDOR_THREAD_NAME) + .build())); + } + + /** + * Starts the vending of backend worker metadata, passing it to endpointsConsumer for processing. + * Returns a {@link CompletableFuture} that can be used to block until the initial {@link + * GetWorkerMetadataStream} has been started. + */ + CompletableFuture<Void> start(Consumer<WindmillEndpoints> endpointsConsumer) { + if (isRunning.compareAndSet(false, true)) { + LOG.debug("Starting WorkerMetadataVendor..."); + + CountDownLatch initialStreamStarted = new CountDownLatch(1); + + workerMetadataFetcher.execute( + () -> { + boolean isInitialWorkerMetadata = true; + WorkerMetadataResponse initialWorkerMetadata = NO_WORKER_METADATA; + + while (isRunning.get()) { + GetWorkerMetadataStream getWorkerMetadataStream = + getWorkerMetadataStreamFactory.create(initialWorkerMetadata, endpointsConsumer); + LOG.debug( + "Starting GetWorkerMetadataStream w/ metadata version {}.", + initialWorkerMetadata.getMetadataVersion()); + getWorkerMetadataStream.start(); + + if (isInitialWorkerMetadata) { + isInitialWorkerMetadata = false; + initialStreamStarted.countDown(); + } + + // Await stream termination and propagate the most current worker metadata to start + // the next stream. + initialWorkerMetadata = awaitGracefulTermination(getWorkerMetadataStream); + LOG.debug( + "Current GetWorkerMetadataStream terminated. Propagating metadata version {} to the next stream.", + initialWorkerMetadata.getMetadataVersion()); + } + }); + + return CompletableFuture.runAsync( + () -> { + int secondsWaited = 0; + try { + while (!initialStreamStarted.await(10, TimeUnit.SECONDS) && isRunning.get()) { + secondsWaited += 10; + LOG.debug( + "Waited {} seconds for initial GetWorkerMetadataStream to start.", + secondsWaited); + } + } catch (InterruptedException e) { + LOG.debug("Interrupted waiting for initial GetWorkerMetadataStream to start."); + Thread.currentThread().interrupt(); + } + }, + // Run this on the calling thread. + MoreExecutors.directExecutor()); + } + + return CompletableFuture.completedFuture(null); + } + + void stop() { + if (isRunning.compareAndSet(true, false)) { + LOG.debug("Shutting down WorkerMetadataVendor..."); + workerMetadataFetcher.shutdownNow(); + boolean isShutdown = false; + try { + isShutdown = + workerMetadataFetcher.awaitTermination( + GET_WORKER_METADATA_STREAM_TIMEOUT_MINUTES, TimeUnit.MINUTES); + } catch (InterruptedException e) { + Thread.currentThread().interrupt(); + } + + if (!isShutdown) { + LOG.warn("Unable to shutdown WorkerMetadataFetcher."); + } + } + } + + /** + * Terminate the stream after {@code GET_WORKER_METADATA_STREAM_TIMEOUT_MINUTES}, and return the + * current WorkerMetadata at the time of termination used to start the next {@link + * GetWorkerMetadataStream}. + */ + private WorkerMetadataResponse awaitGracefulTermination( + GetWorkerMetadataStream getWorkerMetadataStream) { + try { + // Reconnect every now and again to avoid hitting gRPC deadlines. If at any point the server + // closes the stream, we will reconnect immediately; otherwise shutdown the stream after some + // time and create a new one. + if (!getWorkerMetadataStream.awaitTermination( + GET_WORKER_METADATA_STREAM_TIMEOUT_MINUTES, TimeUnit.MINUTES)) { + getWorkerMetadataStream.halfClose(); + } + + // If halfClose() didn't gracefully close the stream, forcefully shutdown(). + if (!getWorkerMetadataStream.awaitTermination(30, TimeUnit.SECONDS)) { + getWorkerMetadataStream.shutdown(); + } + + } catch (InterruptedException e) { + LOG.debug("WorkerMetadataVendor interrupted unexpectedly."); Review Comment: doesn't seem unexpected if stop is calling it. maybe better to not catch here and instead move the try catch to the overall logic in start and use that to break out of the retry loop? -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: github-unsubscr...@beam.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
