kl0u commented on a change in pull request #11460: [FLINK-16655][FLINK-16657]
Introduce embedded executor and use it for Web Submission
URL: https://github.com/apache/flink/pull/11460#discussion_r401898001
##########
File path:
flink-runtime-web/src/main/java/org/apache/flink/runtime/webmonitor/handlers/JarRunHandler.java
##########
@@ -67,41 +69,39 @@ public JarRunHandler(
final MessageHeaders<JarRunRequestBody,
JarRunResponseBody, JarRunMessageParameters> messageHeaders,
final Path jarDir,
final Configuration configuration,
- final Executor executor) {
+ final Executor executor,
+ final Supplier<ApplicationRunner>
applicationRunnerSupplier) {
super(leaderRetriever, timeout, responseHeaders,
messageHeaders);
this.jarDir = requireNonNull(jarDir);
this.configuration = requireNonNull(configuration);
this.executor = requireNonNull(executor);
+
+ this.applicationRunner = applicationRunnerSupplier.get();
}
@Override
protected CompletableFuture<JarRunResponseBody> handleRequest(
@Nonnull final HandlerRequest<JarRunRequestBody,
JarRunMessageParameters> request,
@Nonnull final DispatcherGateway gateway) throws
RestHandlerException {
- final JarHandlerContext context =
JarHandlerContext.fromRequest(request, jarDir, log);
-
- final SavepointRestoreSettings savepointRestoreSettings =
getSavepointRestoreSettings(request);
-
- final CompletableFuture<JobGraph> jobGraphFuture =
getJobGraphAsync(context, savepointRestoreSettings);
-
- CompletableFuture<Integer> blobServerPortFuture =
gateway.getBlobServerPort(timeout);
- CompletableFuture<JobGraph> jarUploadFuture =
jobGraphFuture.thenCombine(blobServerPortFuture, (jobGraph, blobServerPort) -> {
- final InetSocketAddress address = new
InetSocketAddress(gateway.getHostname(), blobServerPort);
- try {
-
ClientUtils.extractAndUploadJobGraphFiles(jobGraph, () -> new
BlobClient(address, configuration));
- } catch (FlinkException e) {
- throw new CompletionException(e);
- }
+ configuration.set(DeploymentOptions.ATTACHED, false);
+ configuration.set(DeploymentOptions.TARGET,
EmbeddedExecutor.NAME);
- return jobGraph;
- });
-
- CompletableFuture<Acknowledge> jobSubmissionFuture =
jarUploadFuture.thenCompose(jobGraph -> gateway.submitJob(jobGraph, timeout));
-
- return jobSubmissionFuture
- .thenCombine(jarUploadFuture, (ack, jobGraph) -> new
JarRunResponseBody(jobGraph.getJobID()));
+ final JarHandlerContext context =
JarHandlerContext.fromRequest(request, jarDir, log);
+ context.applyOnConfiguration(configuration);
+
SavepointRestoreSettings.toConfiguration(getSavepointRestoreSettings(request),
configuration);
+
+ final PackagedProgram program =
context.toPackagedProgram(configuration);
+
+ return CompletableFuture
+ .supplyAsync(() ->
applicationRunner.run(gateway, program, configuration), executor)
+ .thenApply(jobIds -> {
+ if (jobIds.isEmpty()) {
+ throw new
CompletionException(new ProgramInvocationException("No jobs submitted."));
+ }
+ return new
JarRunResponseBody(jobIds.get(0));
Review comment:
Good point @TisonKun .
I agree, that now it is possible to have multiple jobs being executed, but
previously this was not possible because we were throwing an exception at the
first `execute()` in order to extract the `JobGraph`. This is the reason I left
it like this here.
I could either document this behaviour here, or change the
`JarRunResponseBody` to include multiple `JobIds` (which I find more
consistent).
I also include @aljoscha in the discussion to have more opinions.
----------------------------------------------------------------
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]
With regards,
Apache Git Services
