apache/cassandra · error · IllegalStateException

No holder claimed

Error message

No holder claimed 

What it means

CompactionStrategyManager partitions a table's SSTables among holders (repaired/unrepaired, transient, pending-repair, and per-directory strategy holders). getHolderIndex(sstable) is an internal lookup that assumes every SSTable belongs to exactly one holder; if none claims it the internal bookkeeping is inconsistent, so an IllegalStateException is thrown. This is an invariant violation, not user error per se.

Source

Thrown at src/java/org/apache/cassandra/db/compaction/CompactionStrategyManager.java:936

    /**
     * Should only be called holding the readLock
     */
    private void handleFlushNotification(Iterable<SSTableReader> added)
    {
        for (SSTableReader sstable : added)
            getHolder(sstable).addSSTable(sstable);
    }

    private int getHolderIndex(SSTableReader sstable)
    {
        for (int i = 0; i < holders.size(); i++)
        {
            if (holders.get(i).managesSSTable(sstable))
                return i;
        }

        throw new IllegalStateException("No holder claimed " + sstable);
    }

    private AbstractStrategyHolder getHolder(SSTableReader sstable)
    {
        for (AbstractStrategyHolder holder : holders)
        {
            if (holder.managesSSTable(sstable))
                return holder;
        }

        throw new IllegalStateException("No holder claimed " + sstable);
    }

    private AbstractStrategyHolder getHolder(long repairedAt, TimeUUID pendingRepair, boolean isTransient)
    {
        return getHolder(repairedAt != ActiveRepairService.UNREPAIRED_SSTABLE,
                         pendingRepair != ActiveRepairService.NO_PENDING_REPAIR,
                         isTransient);

View on GitHub (pinned to 88fd0f6a0e)

Solutions

1. Restart the node - holders and disk boundaries are rebuilt on startup, reclassifying orphaned sstables
2. Run nodetool relocate / nodetool garbagcollect or trigger a strategy reload (nodetool reloadcompactionstrategy) to resync holder lists
3. Check for concurrent disk-boundary changes and ensure compaction operations are not racing a topology/directory change
4. If reproducible, file a Cassandra bug with the sstable descriptor and compaction strategy config

// before sstable operations, confirm the sstable is managed
List<AbstractStrategyHolder> holders = csm.getHolders();
boolean managed = holders.stream().anyMatch(h -> h.managesSSTable(sstable));
if (!managed) throw new IllegalStateException("sstable not claimed by any holder: " + sstable);

try { idx = csm.getHolderIndex(sstable); }
catch (IllegalStateException e) {
  logger.error("holder bookkeeping out of sync; restart/reload strategy", e);
  cfs.getCompactionStrategyManager().maybeReloadDiskBoundaries();
}

When it happens

Trigger: An SSTableReader is passed to getHolderIndex/getHolder (e.g. via compaction task creation, sstable lifecycle operations, or nodetool sstable operations) whose location (directory index, repaired/pending-repair/transient state) does not match any registered holder - typically because disk boundaries or strategy state changed without the sstable being reclassified.

Common situations: Race between disk-boundary relocation/reload and compaction on the same sstable; sstables moved between directories out-of-band; bugs in pending-repair lifecycle (session finished/forgotten while sstable still tagged); running with JBOD and changing data directories.

Understand the failure class

Background: "This is a bug, please report it": internal invariant violations, unreachable panics, and SNH errors explained — this error's family across 47 libraries.

Related errors

• No holder claimed isPendingRepair: %s, isPendingRepair %s
• You can't mix repaired and unrepaired data in a compaction
• Could not reference sstables
• Failed to create a key iterator for sstable ${filename}
• Last written key %s >= current key %s, writing into %s

AI-assisted analysis of apache/cassandra@88fd0f6a0e (2026-09-10). Data as JSON: /api/errors/b4dbdfe5b428911b. Report an issue: GitHub.