Cassandra

Metrics collection

Node-Level

Configuration data

  • Version
  • Cluster
  • Datacenter
  • Rack
  • Host ID
  • Node Status
  • Gossip, Thrift, CQL/Native Status
  • Tokens

Performance metrics

Read Requests

Number of read requests is collected from org.apache.cassandra.metrics:type=ClientRequest, scope=Read,name=Latency#Count.

Granularity: 1 second.

Write Requests

Number of write requests is collected from org.apache.cassandra.metrics:type=ClientRequest, scope=Write,name=Latency#Count.

Granularity: 1 second.

Client Read Latencies

Mean, 50th, 95th and 99th percentile are collected from org.apache.cassandra.metrics:type=ClientRequest, scope=Read,name=Latency.

Granularity: 1 second.

Client Write Latencies

Mean, 50th, 95th and 99th percentile are collected from org.apache.cassandra.metrics:type=ClientRequest, scope=Write,name=Latency.

Granularity: 1 second.

Pending Requests

Pending requests in threadpools are collected from org.apache.cassandra.metrics:type=ThreadPools,path=request, scope=CounterMutationStage,name=PendingTasks.

Granularity: 1 second.

Blocked Requests

Blocked requests in threadpools org.apache.cassandra.metrics:type=ThreadPools,path=request, scope=CounterMutationStage,name=CurrentlyBlockedTasks.

Granularity: 1 second.

Dropped Messages

Write (Mutation), read, counter mutation, read repair, and request/response are collected from the dropped messages field of the MessagingService instance.

Granularity: 1 second.

Keyspaces

Reads/writes, average read/write latency, SSTables, and disk size are collected from the schema instance and ColumnFamilyStore per keyspace.

Granularity: 1 second.

Compactions

Pending compactions is collected from the pending compactions field of the CompactionManager.

Granularity: 1 second.

Cache Hits

Counter, key, and row cache are collected from org.apache.cassandra.metrics:type=Cache,name=HitRate.

Granularity: 1 second.

Bloom Filter

Bloom filter miss rate is collected from org.apache.cassandra.metrics:type=ColumnFamily,name=BloomFilterFalseRatio.

Granularity: 1 second.

Health signatures

For each sensor, there is a curated knowledgebase of health signatures that are evaluated continuously against the incoming metrics and are used to raise issues or incidents depending on user impact.

Built-in events trigger issues or incidents based on failing health signatures on entities, and custom events trigger issues or incidents based on the thresholds of an individual metric of any given entity.

For information about built-events for Cassandra Node, see the Built-in events reference.

Cluster-level

Configuration data

  • Name
  • Available Nodes
  • Unreachable Nodes
  • Keyspaces

Performance metrics

Data point Description Granularity
Overall Requests Sum of all the requests for Cassandra nodes. 1 second
Client Latencies Average client read and write latencies. 1 second
Overall Disk Size Sum of all the disk sizes for Cassandra nodes. 1 second
Replication Factors Replication factor per keyspace. 1 second
Disk Sizes Disk size per keyspace. 1 second

Health signatures

For information about built-events for Cassandra Cluster, see the Built-in events reference.