Addressing Mutation Thread Stalls in Cassandra 4.1.8

At NetApp Instaclustr, we’re committed to helping our customers run stable and performant Cassandra clusters. As part of that commitment, we want to share an important update regarding Apache Cassandra 4.1.8.

Issue Details

In June 2025, NetApp began investigating a recurring issue on the NetApp Instaclustr platform that has been observed on Apache Cassandra version 4.1.8. This issue was observed to cause mutation threads responsible for handling write operations to stall under certain conditions. When this occurs, usually during high write throughput or uneven memory pressure, write operations may cease to progress and remain blocked until the affected node is restarted. This behaviour was observed across multiple clusters running Cassandra 4.1.8 on the NetApp Instaclustr platform.

Issue Details

Symptom Impact Resolution
Mutation threads stall in MemtableAllocator, causing pending writes to accumulate. Increased read/write latency, dropped messages, and node restarts required to recover. Issue resolved in Cassandra 4.1.9. See https://issues.apache.org/jira/browse/CASSANDRA-20084.

Issue Analysis

Clusters affected by this issue exhibit a consistent set of symptoms that indicate a stall in the mutation thread processing pipeline:

  • MutationStage threads enter a WAITING state.
  • Completed MutationStages drop to zero, indicating halted progress.
  • Pending MutationStages increase steadily, suggesting an accumulation of write operations.
  • Active MutationStages plateau at the maximum thread count, with no new threads being processed.
  • Flushing does not resolve the issue; recovery requires a full node restart.

Mitigation

Managed Platform Customers

The issue is resolved in Apache Cassandra 4.1.9, which is now available on our platform. Cassandra 4.1.9 also includes additional stability and security improvements. See the full release notes here: https://github.com/apache/cassandra/blob/cassandra-4.1.9/CHANGES.txt

Next steps:

  • All clusters currently running Cassandra 4.1.8 will be upgraded to 4.1.9.
  • Customers will receive a scheduled maintenance notification prior to their upgrade.
  • As part of this upgrade, no action is required from your side. Applications running on Cassandra 4.1.8 will be fully compatible with Cassandra 4.1.9.

Enterprise Support Customers

Please upgrade your clusters running on Cassandra 4.1.8 to Cassandra 4.1.9 or a later version to avoid potential mutation thread stalls.

Lifecycle Status

Guidance for our managed service and enterprise Support customers

Please use Cassandra 4.1.9 or later for all new deployments as soon as is practical.

 

If you have any further queries regarding this, please contact Instaclustr Support.