apache/cassandra · warning · RuntimeException

Unable to cancel in-progress compactions for

Error message

Unable to cancel in-progress compactions for 

What it means

CompactionManager.forceCompaction (and similar forced-compaction paths) must first pause in-flight compactions via ColumnFamilyStore.runWithCompactionsDisabled. That helper returns null when it fails to obtain the compaction lock (i.e. it could not cancel/disable running compactions within its wait window). Cassandra throws this RuntimeException rather than silently proceeding, because running a forced compaction concurrently with in-flight compactions would corrupt lifecycle assumptions.

Source

Thrown at src/java/org/apache/cassandra/db/compaction/CompactionManager.java:1252

        Callable<CompactionTasks> taskCreator = () -> {
            Collection<SSTableReader> sstables = sstablesFn.get();
            if (sstables == null || sstables.isEmpty())
            {
                logger.debug("No sstables found for the provided token range");
                return CompactionTasks.empty();
            }
            return cfStore.getCompactionStrategyManager().getUserDefinedTasks(sstables, cfStore.getDefaultGcBefore(FBUtilities.nowInSeconds()));
        };

        try (CompactionTasks tasks = cfStore.runWithCompactionsDisabled(taskCreator,
                                                                        sstablesPredicate,
                                                                        OperationType.MAJOR_COMPACTION,
                                                                        false,
                                                                        false,
                                                                        false))
        {
            if (tasks == null)
                throw new RuntimeException("Unable to cancel in-progress compactions for " + cfStore.getKeyspaceName() + '.' + cfStore.getTableName() + ". Usually retrying will work.");

            if (tasks.isEmpty())
                return;

            Runnable runnable = new WrappedRunnable()
            {
                protected void runMayThrow()
                {
                    for (AbstractCompactionTask task : tasks)
                        if (task != null)
                        {
                            task.setCompactionType(OperationType.MAJOR_COMPACTION);
                            task.execute(active);
                        }
                }
            };

            FBUtilities.waitOnFuture(executor.submitIfRunning(runnable, "force compaction for token range"));

View on GitHub (pinned to 88fd0f6a0e)

Solutions

1. Retry the operation after a short delay - the message itself notes 'Usually retrying will work'
2. Reduce concurrent compaction pressure (lower concurrent_compactors or compaction_throughput is not needed; wait for backlog to shrink)
3. Use nodetool stop COMPACTION to halt current compactions, then retry
4. Schedule major compactions during low-write windows

Example fix

// before
forceCompaction(cfStore, sstablesFn, predicate); // throws when compactions busy
// after
try { forceCompaction(cfStore, sstablesFn, predicate); }
catch (RuntimeException e) { if (e.getMessage().startsWith("Unable to cancel")) { Uninterruptibles.sleepUninterruptibly(5, TimeUnit.SECONDS); retry(); } else throw e; }

// check compaction activity first via JMX / StorageService
boolean busy = !CompactionManager.instance.getCompactions().isEmpty();
if (busy) throw new IllegalStateException("compactions in progress; defer major compaction");

try { forceCompaction(cfs, fn, pred); }
catch (RuntimeException e) {
  if (e.getMessage() != null && e.getMessage().startsWith("Unable to cancel in-progress compactions")) {
    Uninterruptibles.sleepUninterruptibly(10, TimeUnit.SECONDS);
    // retry with bounded attempts
  } else throw e;
}

When it happens

Trigger: Calling forceCompaction / nodetool-triggered major compaction (or compact with token ranges) on a table that has active compactions that cannot be cancelled within runWithCompactionsDisabled's timeout; the returned CompactionTasks is null.

Common situations: Tables under heavy continuous write load with constant background compaction; large compaction storms after bulk loading; issuing major compaction while incremental repair or anticompaction is running; repeated compact requests on a busy node.

Understand the failure class

Background: "Invalid state transition" errors: "status must be X, actually Y", "already rejected/charging/uninstalled", "cannot ... while running" — what they mean when a library rejects your call — this error's family across 31 libraries.

Related errors

• Unable to cancel in-progress compactions for {keyspace}.{tab
• concurrent_compactors should be strictly greater than 0, but
• Could not set new local compaction strategy: <cause message>
• The min_compaction_threshold cannot be larger than the max_c
• Disabling compaction by setting min_compaction_threshold or

AI-assisted analysis of apache/cassandra@88fd0f6a0e (2026-09-10). Data as JSON: /api/errors/31a4e75be2683ff6. Report an issue: GitHub.