chamikaramj commented on a change in pull request #13605:
URL: https://github.com/apache/beam/pull/13605#discussion_r550936568
##########
File path:
runners/google-cloud-dataflow-java/src/main/java/org/apache/beam/runners/dataflow/DataflowRunner.java
##########
@@ -1186,6 +1190,45 @@ public DataflowPipelineJob run(Pipeline pipeline) {
return dataflowPipelineJob;
}
+ static void configureSdkHarnessContainerImages(
+ DataflowPipelineOptions options,
+ RunnerApi.Pipeline pipelineProto,
+ Job newJob,
+ String workerHarnessContainerImage) {
+ if (hasExperiment(options, "use_runner_v2") || hasExperiment(options,
"use_unified_worker")) {
+ ImmutableSet.Builder<String> sdkContainerUrlSetBuilder =
ImmutableSet.builder();
+ sdkContainerUrlSetBuilder.add(workerHarnessContainerImage);
+ for (Map.Entry<String, RunnerApi.Environment> entry :
+ pipelineProto.getComponents().getEnvironmentsMap().entrySet()) {
+ if
(!BeamUrns.getUrn(RunnerApi.StandardEnvironments.Environments.DOCKER)
+ .equals(entry.getValue().getUrn())) {
+ throw new RuntimeException(
+ "Dataflow can only execute pipeline steps in Docker
environments: "
+ + entry.getValue().getUrn());
+ }
+ RunnerApi.DockerPayload dockerPayload;
+ try {
+ dockerPayload =
RunnerApi.DockerPayload.parseFrom(entry.getValue().getPayload());
+ } catch (InvalidProtocolBufferException e) {
+ throw new RuntimeException("Error parsing docker payload.", e);
+ }
+ sdkContainerUrlSetBuilder.add(dockerPayload.getContainerImage());
Review comment:
It might make sense to de-dupe here in-case same Docker URL gets added
twice (we can generalize this when Dataflow supports multiple environments with
the same Docker URL hence please add a TODO).
##########
File path:
runners/google-cloud-dataflow-java/src/main/java/org/apache/beam/runners/dataflow/DataflowRunner.java
##########
@@ -1186,6 +1190,45 @@ public DataflowPipelineJob run(Pipeline pipeline) {
return dataflowPipelineJob;
}
+ static void configureSdkHarnessContainerImages(
+ DataflowPipelineOptions options,
+ RunnerApi.Pipeline pipelineProto,
+ Job newJob,
+ String workerHarnessContainerImage) {
+ if (hasExperiment(options, "use_runner_v2") || hasExperiment(options,
"use_unified_worker")) {
+ ImmutableSet.Builder<String> sdkContainerUrlSetBuilder =
ImmutableSet.builder();
+ sdkContainerUrlSetBuilder.add(workerHarnessContainerImage);
+ for (Map.Entry<String, RunnerApi.Environment> entry :
+ pipelineProto.getComponents().getEnvironmentsMap().entrySet()) {
+ if
(!BeamUrns.getUrn(RunnerApi.StandardEnvironments.Environments.DOCKER)
+ .equals(entry.getValue().getUrn())) {
+ throw new RuntimeException(
+ "Dataflow can only execute pipeline steps in Docker
environments: "
+ + entry.getValue().getUrn());
+ }
+ RunnerApi.DockerPayload dockerPayload;
+ try {
+ dockerPayload =
RunnerApi.DockerPayload.parseFrom(entry.getValue().getPayload());
+ } catch (InvalidProtocolBufferException e) {
+ throw new RuntimeException("Error parsing docker payload.", e);
+ }
+ sdkContainerUrlSetBuilder.add(dockerPayload.getContainerImage());
+ }
+ List<SdkHarnessContainerImage> sdkContainerList =
Review comment:
Do we need any special handling for pipeline SDK similar to following
for Python ?
https://github.com/apache/beam/blob/master/sdks/python/apache_beam/runners/dataflow/internal/apiclient.py#L285
##########
File path:
runners/google-cloud-dataflow-java/src/test/java/org/apache/beam/runners/dataflow/DataflowRunnerTest.java
##########
@@ -1342,6 +1347,57 @@ public void testTransformTranslator() throws IOException
{
assertTrue(transform.translated);
}
+ @Test
+ public void testSdkHarnessConfiguration() throws IOException {
Review comment:
Please also try out a Dataflow GCE job (or existing ITs).
##########
File path:
runners/google-cloud-dataflow-java/src/main/java/org/apache/beam/runners/dataflow/DataflowRunner.java
##########
@@ -1186,6 +1190,45 @@ public DataflowPipelineJob run(Pipeline pipeline) {
return dataflowPipelineJob;
}
+ static void configureSdkHarnessContainerImages(
+ DataflowPipelineOptions options,
+ RunnerApi.Pipeline pipelineProto,
+ Job newJob,
+ String workerHarnessContainerImage) {
+ if (hasExperiment(options, "use_runner_v2") || hasExperiment(options,
"use_unified_worker")) {
+ ImmutableSet.Builder<String> sdkContainerUrlSetBuilder =
ImmutableSet.builder();
+ sdkContainerUrlSetBuilder.add(workerHarnessContainerImage);
+ for (Map.Entry<String, RunnerApi.Environment> entry :
+ pipelineProto.getComponents().getEnvironmentsMap().entrySet()) {
+ if
(!BeamUrns.getUrn(RunnerApi.StandardEnvironments.Environments.DOCKER)
+ .equals(entry.getValue().getUrn())) {
+ throw new RuntimeException(
+ "Dataflow can only execute pipeline steps in Docker
environments: "
+ + entry.getValue().getUrn());
+ }
+ RunnerApi.DockerPayload dockerPayload;
+ try {
+ dockerPayload =
RunnerApi.DockerPayload.parseFrom(entry.getValue().getPayload());
+ } catch (InvalidProtocolBufferException e) {
+ throw new RuntimeException("Error parsing docker payload.", e);
+ }
+ sdkContainerUrlSetBuilder.add(dockerPayload.getContainerImage());
+ }
+ List<SdkHarnessContainerImage> sdkContainerList =
+ sdkContainerUrlSetBuilder.build().stream()
+ .map(
+ (String url) -> {
+ SdkHarnessContainerImage image = new
SdkHarnessContainerImage();
+ image.setContainerImage(url);
+ return image;
+ })
+ .collect(Collectors.toList());
+ for (WorkerPool workerPool : newJob.getEnvironment().getWorkerPools()) {
+ workerPool.setSdkHarnessContainerImages(sdkContainerList);
Review comment:
For any Python environments, we should set 'useSingleCorePerContainer'
to true for efficiency.
##########
File path:
runners/google-cloud-dataflow-java/src/main/java/org/apache/beam/runners/dataflow/DataflowRunner.java
##########
@@ -1186,6 +1190,45 @@ public DataflowPipelineJob run(Pipeline pipeline) {
return dataflowPipelineJob;
}
+ static void configureSdkHarnessContainerImages(
+ DataflowPipelineOptions options,
+ RunnerApi.Pipeline pipelineProto,
+ Job newJob,
+ String workerHarnessContainerImage) {
+ if (hasExperiment(options, "use_runner_v2") || hasExperiment(options,
"use_unified_worker")) {
Review comment:
Is there a util function that we can depend on (similar to [1] for
Python) introduce of checking these properties separately for runner v2 which
could lead to inconsistencies ?
[1]
https://github.com/apache/beam/blob/master/sdks/python/apache_beam/runners/dataflow/internal/apiclient.py#L1039
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
[email protected]
