apache/cassandra · error · java.lang.IllegalStateException

Can not commit transformation: "%s"(%s).

Error message

Can not commit transformation: "%s"(%s).

What it means

failureHandler builds the CommitFailureHandler used by decommission and move. When committing a topology transformation (e.g. PrepareMove or the decommission prepare) fails, it marks the operation failed (e.g. StorageService.markMoveFailed), logs, and throws IllegalStateException carrying the commit result code and message.

Source

Thrown at src/java/org/apache/cassandra/tcm/sequences/SingleNodeSequences.java:207

        ClusterMetadataService.instance().commit(new PrepareMove(self,
                                                                 Collections.singleton(newToken),
                                                                 ClusterMetadataService.instance().placementProvider(),
                                                                 true),
                                                 m -> m,
                                                 failureHandler("PrepareMove", StorageService.instance::markMoveFailed));
        InProgressSequences.finishInProgressSequences(self);

        if (logger.isDebugEnabled())
            logger.debug("Successfully moved to new token {}", StorageService.instance.getLocalTokens().iterator().next());
    }

    private static ClusterMetadataService.CommitFailureHandler<ClusterMetadata> failureHandler(String type, Runnable markFailed)
    {
        return (code, msg) -> {
            logger.warn("Got failure committing {} transformation: {} {}", type, code, msg);
            markFailed.run();
            throw new IllegalStateException(String.format("Can not commit transformation: \"%s\"(%s).", code, msg));
        };
    }

    static void resumeMove()
    {
        if (ClusterMetadataService.instance().isMigrating() || ClusterMetadataService.state() == ClusterMetadataService.State.GOSSIP)
            throw new IllegalStateException("This cluster is migrating to cluster metadata, can't move until that is done.");

        ClusterMetadata metadata = ClusterMetadata.current();
        NodeId self = metadata.myNodeId();
        MultiStepOperation<?> sequence = metadata.inProgressSequences.get(self);
        if (sequence == null || sequence.kind() != MultiStepOperation.Kind.MOVE)
        {
            String msg = "No move operation in progress, can't resume";
            logger.info(msg);
            if (StorageService.instance.operationMode() == MOVE_FAILED)
            {
                // there is no ongoing move to resume, but operation mode thinks there is

View on GitHub (pinned to 88fd0f6a0e)

Solutions

1. Check the logged code+message ('Got failure committing ... transformation') for the root cause and fix it (restore CMS quorum, resolve concurrent operation).
2. Verify CMS availability and connectivity, then resume the operation: `nodetool resumemove` / finish the decommission.
3. If the operation cannot proceed, abort it (`nodetool abortmove` / `abortdecommission`) and retry later.
4. Ensure no other topology operation is in flight that conflicts with this one.

if (!ClusterMetadataService.instance().isCurrent(CMSId)) return; // ensure CMS reachable before committing topology ops

try { move(token); }
catch (IllegalStateException e) { if (e.getMessage().startsWith("Can not commit transformation")) { resumeOrAbort(); } else throw e; }

When it happens

Trigger: ClusterMetadataService.commit() invoking the failure handler for a PrepareMove/PrepareLeave transformation - typically because the CMS rejected or could not replicate the transformation (quorum loss, rejected epoch, precondition failure).

Common situations: CMS majority unavailable during a move/decommission; transformation rejected due to concurrent topology change; network partition between the node and CMS members.

Related errors

• Unable to commit rack changes: {r}
• Can not remove a node that has an in-progress sequence
• This cluster is migrating to cluster metadata, can't move un
• Unknown endpoint
• Cannot return topology when accord.enabled = false in cassan

AI-assisted analysis of apache/cassandra@88fd0f6a0e (2026-09-10). Data as JSON: /api/errors/46aaf47e3166ea14. Report an issue: GitHub.