apache/cassandra · warning · RuntimeException

Unexpected interruption

Error message

Unexpected interruption

What it means

In StressAction's inner run loop, each worker thread waits on a 'start' CountDownLatch so all threads begin simultaneously. If the worker thread is interrupted while waiting on start.await(), the code rethrows as RuntimeException('Unexpected interruption', e). Stress threads are not expected to be interrupted during startup; this means the JVM/harness interrupted the thread abnormally (e.g. shutdown hook, external kill semantics, or a bug in the coordinating code).

Source

Thrown at tools/stress/src/org/apache/cassandra/stress/StressAction.java:254

        final Consumer[] consumers = new Consumer[threadCount];
        for (int i = 0; i < threadCount; i++)
        {
            consumers[i] = new Consumer(operations, isWarmup,
                                        done, start, releaseConsumers, workManager, metrics, rateLimiter);
        }

        // starting worker threadCount
        for (int i = 0; i < threadCount; i++)
            consumers[i].start();

        // wait for the lot of them to get their pants on
        try
        {
            start.await();
        }
        catch (InterruptedException e)
        {
            throw new RuntimeException("Unexpected interruption", e);
        }
        // start counting from NOW!
        if(rateLimiter != null)
        {
            rateLimiter.start();
        }
        // release the hounds!!!
        releaseConsumers.countDown();

        metrics.start();

        if (durationUnits != null)
        {
            Uninterruptibles.sleepUninterruptibly(duration, durationUnits);
            workManager.stop();
        }
        else if (opCount <= 0)
        {

View on GitHub (pinned to 88fd0f6a0e)

Solutions

1. Check who interrupts the worker (shutdown hooks, cancel paths) and ensure stress is not torn down during startup.
2. Re-run the stress command and avoid killing the process during the first seconds of the run.
3. If interruption is intentional (cancel), treat this RuntimeException as expected cancellation in the harness.
4. Wrap the StressAction invocation with try/catch (RuntimeException) checking getCause() instanceof InterruptedException for graceful shutdown.

Example fix

// before
stressAction.run(); // process may die with 'Unexpected interruption'
// after
try {
    stressAction.run();
} catch (RuntimeException e) {
    if (!(e.getCause() instanceof InterruptedException)) throw e;
    // expected during shutdown
}

// ensure no component interrupts the stress thread during coordinated start:
// avoid calling stressThread.interrupt() before start.countDown() fires

try {
    start.await();
} catch (InterruptedException e) {
    Thread.currentThread().interrupt(); // restore interrupt status
    throw new RuntimeException("Unexpected interruption", e);
}

When it happens

Trigger: Thread interrupt delivered while a stress worker is blocked in start.await() before the coordinated start — e.g. JVM shutdown initiated, StressAction cancel/interrupt path racing with worker start, or a misconfigured pool shutting down workers.

Common situations: Ctrl-C or timeout killing the stress process just as threads start; a supervising harness calling interrupt on the stress thread pool; bug fixes/patches that interrupt workers during setup.

Related errors

• Queue is empty
• UnsupportedOperationException
• Maximum pool size has been changed while resizing
• Must be one of ${values}
• ${executor} has shut down

AI-assisted analysis of apache/cassandra@88fd0f6a0e (2026-09-10). Data as JSON: /api/errors/4bec4b100f7e7fc4. Report an issue: GitHub.