Troubleshooting

As any distributed database does, sometimes Cassandra breaks and you will have to troubleshoot what is going on. Generally speaking you can debug Cassandra like any other distributed Java program, meaning that you have to find which machines in your cluster are misbehaving and then isolate the problem using logs and tools. Luckily Cassandra had a great set of instrospection tools to help you.

These pages include a number of command examples demonstrating various debugging and analysis techniques, mostly for Linux/Unix systems. If you don’t have access to the machines running Cassandra, or are running on Windows or another operating system you may not be able to use the exact commands but there are likely equivalent tools you can use.

• Find The Misbehaving Nodes
  • Client Logs and Errors
  • Metrics
  • Next Step: Investigate the Node(s)
• Cassandra Logs
  • Common Log Files
  • Getting More Information
• Use Nodetool
  • Cluster Status
  • Coordinator Query Latency
  • Local Query Latency
  • Threadpool State
  • Compaction State
• Diving Deep, Use External Tools
  • JVM Tooling
  • Basic OS Tooling
  • Advanced tools