Examples with RESTARTING com.hazelcast.jet.core.JobStatus.RESTARTING used on opensource projects

Example 1 with RESTARTING

use of com.hazelcast.jet.core.JobStatus.RESTARTING in project hazelcast-jet by hazelcast.

the class MasterContext method invokeCompleteExecution.

private void invokeCompleteExecution(Throwable error) {
    JobStatus status = jobStatus();
    Throwable finalError;
    if (status == STARTING || status == RESTARTING || status == RUNNING) {
        logger.fine("Completing " + jobIdString());
        finalError = error;
    } else {
        if (error != null) {
            logger.severe("Cannot properly complete failed " + jobIdString() + ": status is " + status, error);
        } else {
            logger.severe("Cannot properly complete " + jobIdString() + ": status is " + status);
        }
        finalError = new IllegalStateException("Job coordination failed.");
    }
    Function<ExecutionPlan, Operation> operationCtor = plan -> new CompleteExecutionOperation(executionId, finalError);
    invoke(operationCtor, responses -> finalizeJob(error), null);
}

Example 2 with RESTARTING

use of com.hazelcast.jet.core.JobStatus.RESTARTING in project hazelcast-jet by hazelcast.

the class SplitBrainTest method when_quorumIsLostOnBothSides_then_jobRestartsUntilMerge.

@Test
public void when_quorumIsLostOnBothSides_then_jobRestartsUntilMerge() {
    int firstSubClusterSize = 2;
    int secondSubClusterSize = 2;
    int clusterSize = firstSubClusterSize + secondSubClusterSize;
    StuckProcessor.executionStarted = new CountDownLatch(clusterSize * PARALLELISM);
    Job[] jobRef = new Job[1];
    Consumer<JetInstance[]> beforeSplit = instances -> {
        MockPS processorSupplier = new MockPS(StuckProcessor::new, clusterSize);
        DAG dag = new DAG().vertex(new Vertex("test", processorSupplier));
        jobRef[0] = instances[0].newJob(dag, new JobConfig().setSplitBrainProtection(true));
        assertOpenEventually(StuckProcessor.executionStarted);
    };
    BiConsumer<JetInstance[], JetInstance[]> onSplit = (firstSubCluster, secondSubCluster) -> {
        StuckProcessor.proceedLatch.countDown();
        long jobId = jobRef[0].getId();
        assertTrueEventually(() -> {
            JetService service1 = getJetService(firstSubCluster[0]);
            JetService service2 = getJetService(secondSubCluster[0]);
            assertEquals(RESTARTING, service1.getJobCoordinationService().getJobStatus(jobId));
            assertEquals(STARTING, service2.getJobCoordinationService().getJobStatus(jobId));
        });
        assertTrueAllTheTime(() -> {
            JetService service1 = getJetService(firstSubCluster[0]);
            JetService service2 = getJetService(secondSubCluster[0]);
            assertEquals(RESTARTING, service1.getJobCoordinationService().getJobStatus(jobId));
            assertEquals(STARTING, service2.getJobCoordinationService().getJobStatus(jobId));
        }, 20);
    };
    Consumer<JetInstance[]> afterMerge = instances -> {
        assertTrueEventually(() -> {
            assertEquals(clusterSize * 2, MockPS.initCount.get());
            assertEquals(clusterSize * 2, MockPS.closeCount.get());
        });
        assertEquals(clusterSize, MockPS.receivedCloseErrors.size());
        MockPS.receivedCloseErrors.forEach(t -> assertTrue(t instanceof TopologyChangedException));
    };
    testSplitBrain(firstSubClusterSize, secondSubClusterSize, beforeSplit, onSplit, afterMerge);
}