apache/flink · error · ProgramInvocationException

Job failed

Error message

Job failed

What it means

Thrown by ClusterClientJobClientAdapter.getJobExecutionResult when jobResult.toJobExecutionResult(classLoader) throws a Throwable. The JobResult object carries the serialized exception (if any) from the job's failure, and toJobExecutionResult re-throws it during deserialization. This wraps the original failure cause in a ProgramInvocationException with message 'Job failed'. This is the user-facing signal that the job did not complete successfully when using the JobClient API.

Source

Thrown at flink-clients/src/main/java/org/apache/flink/client/deployment/ClusterClientJobClientAdapter.java:129

                clusterClientProvider,
                (clusterClient -> clusterClient.getAccumulators(jobID, classLoader)));
    }

    @Override
    public CompletableFuture<JobExecutionResult> getJobExecutionResult() {
        checkNotNull(classLoader);

        return bridgeClientRequest(
                clusterClientProvider,
                (clusterClient ->
                        clusterClient
                                .requestJobResult(jobID)
                                .thenApply(
                                        (jobResult) -> {
                                            try {
                                                return jobResult.toJobExecutionResult(classLoader);
                                            } catch (Throwable t) {
                                                throw new CompletionException(
                                                        new ProgramInvocationException(
                                                                "Job failed", jobID, t));
                                            }
                                        })));
    }

    @Override
    public CompletableFuture<CoordinationResponse> sendCoordinationRequest(
            String operatorUid, CoordinationRequest request) {
        return bridgeClientRequest(
                clusterClientProvider,
                clusterClient ->
                        clusterClient.sendCoordinationRequest(jobID, operatorUid, request));
    }

    @Override
    public void reportHeartbeat(long expiredTimestamp) {
        bridgeClientRequest(

View on GitHub (pinned to 2f3c205e92)

Solutions

1. Inspect the root cause in the ProgramInvocationException to identify the actual job failure reason
2. Check the JobManager logs and the Flink Web UI for the ArchivedExecutionGraph exception
3. Fix the underlying job logic or resource issue, then resubmit
4. Use jobClient.getJobStatus() to poll before calling getJobExecutionResult() if you want to handle failures gracefully

Example fix

// before
JobExecutionResult result = jobClient.getJobExecutionResult().get();

// after
try {
    JobExecutionResult result = jobClient.getJobExecutionResult().get();
} catch (ExecutionException e) {
    if (e.getCause() instanceof ProgramInvocationException) {
        Throwable jobFailure = e.getCause().getCause();
        LOG.error("Job failed with cause:", jobFailure);
    }
    throw e;
}

try {
    JobExecutionResult result = jobClient.getJobExecutionResult().get();
} catch (ExecutionException e) {
    Throwable cause = e.getCause();
    if (cause instanceof ProgramInvocationException) {
        Throwable jobFailure = cause.getCause();
        LOG.error("Job failed due to: {}", jobFailure.getMessage(), jobFailure);
    }
}

When it happens

Trigger: The submitted job finished in a FAILED state due to a runtime exception in user code, a serialization error, or an operator crash. When the JobClient's getJobExecutionResult() future completes, it attempts to materialize the result and encounters the stored exception from the ArchivedExecutionGraph.

Common situations: User code throws an unhandled exception inside a ProcessFunction/MapFunction; OutOfMemoryError during job execution; checkpoint failure that triggers job failure; source/sink connector timeout or connection refused.

AI-assisted analysis of apache/flink@2f3c205e92 (2026-08-14). Data as JSON: /api/errors/222b25be1e172a74. Report an issue: GitHub.