oracle/graal · error · RuntimeException

Interrupted while waiting for housekeeping thread shutdown.

Error message

Interrupted while waiting for housekeeping thread shutdown.

What it means

ObjectPoolingAllocator.shutdown() in LockFreePrefixTree stops the housekeeping thread and joins it; if the calling thread is interrupted while waiting in join(), the InterruptedException is wrapped in a RuntimeException with this message. Shutdown itself completed the disable signal; only the wait-for-exit was cut short, so the housekeeping thread may still be finishing asynchronously.

Source

Thrown at sdk/src/org.graalvm.collections/src/org/graalvm/collections/LockFreePrefixTree.java:835

            } else {
                if (sizeClass >= missedHashChildrenRequestCounts.length()) {
                    throw INTERNAL_FAILURE_EXCEPTION;
                }
                missedHashChildrenRequestCounts.incrementAndGet(sizeClass);
                throw FAILED_ALLOCATION_EXCEPTION;
            }
        }

        /**
         * @since 23.0
         */
        @Override
        public void shutdown() {
            housekeepingThread.isEnabled.set(false);
            try {
                housekeepingThread.join();
            } catch (InterruptedException e) {
                throw new RuntimeException("Interrupted while waiting for housekeeping thread shutdown.", e);
            }
        }

        public String status() {
            StringBuilder content = new StringBuilder();
            content.append("ObjectPoolingAllocator").append(System.lineSeparator());
            content.append("======================").append(System.lineSeparator());

            // Misses statistics.
            content.append("  current node alloc misses:      ").append(missedNodePoolRequestCount.get()).append(System.lineSeparator());
            content.append("  current linear children misses: ").append(System.lineSeparator());
            content.append("    size class ");
            for (int sizeClass = 0; sizeClass < SIZE_CLASS_COUNT; sizeClass++) {
                content.append(String.format("%4d", sizeClass));
            }
            content.append(System.lineSeparator());
            content.append("    miss count ");
            for (int sizeClass = 0; sizeClass < SIZE_CLASS_COUNT; sizeClass++) {

View on GitHub (pinned to a66e9ccd1d)

Solutions

1. Clear or handle the interrupt before calling shutdown(): check and consume Thread.interrupted(), or call shutdown() outside interrupt scope.
2. Catch the RuntimeException and, if needed, re-check the housekeeping thread with a bounded join() after clearing the interrupt flag.
3. Restructure so shutdown() is invoked from a non-interruptible context (dedicated closer thread or before executor shutdown).

Example fix

// before
executor.shutdownNow(); // interrupts tasks
alloc.shutdown(); // inside interrupted task -> RuntimeException

// after
Thread.interrupted(); // clear any pending interrupt flag
alloc.shutdown();

Thread.interrupted(); // clear flag before shutdown
alloc.shutdown();

try {
    alloc.shutdown();
} catch (RuntimeException e) {
    if (e.getCause() instanceof InterruptedException) {
        Thread.currentThread().interrupt(); // restore flag
        // shutdown was signalled; optionally re-join with a bounded wait
    } else {
        throw e;
    }
}

When it happens

Trigger: Calling shutdown() on the allocator while the calling thread's interrupt flag is set (e.g. inside a task cancelled by an ExecutorService, or after Thread.interrupt() from a timeout watchdog). Also when shutdown() races with JVM shutdown hooks that interrupt threads.

Common situations: Running the prefix tree inside thread pools that cancel tasks on timeout; test harnesses that interrupt workers between phases; shutdown hooks racing with executor.shutdownNow(), which interrupts pending tasks.

Related errors

• Allocator's housekeeping thread was interrupted.
• endless collision link cycle, most likely due to unsynchroni
• Key in the prefix tree cannot be 0.
• Cannot signal lock that is not held
• You can't resume an already executing continuation.

AI-assisted analysis of oracle/graal@a66e9ccd1d (2026-08-14). Data as JSON: /api/errors/7c9117fe569adc2d. Report an issue: GitHub.