Cassandra

Auto-Detection

Built-in Metrics

Bleemeo monitors Cassandra through automatic service detection and JMX metrics collection for cluster performance.

Auto-Detection

Glouton automatically detects Cassandra instances running on the system.

To enable metrics gathering, the Bleemeo agent must be installed with JMX enabled. See Java JMX Metrics for setup details.

Cassandra must also expose JMX over a TCP port:

• Docker: Add environment variable JVM_OPTS=-Dcassandra.jmx.remote.port=7199.
• Native package: No additional configuration is needed. Cassandra exposes JMX on port 7199 on localhost by default.
• Other setups: Add -Dcom.sun.management.jmxremote.port=7199 -Dcom.sun.management.jmxremote.authenticate=false to the JVM options.

Caution

This allows unauthenticated access. Ensure no untrusted access to this port is possible, or set up authenticated JMX access.

If auto-detected parameters are incorrect, you can override them manually.

Linux

sudo tee /etc/glouton/conf.d/99-cassandra.conf > /dev/null << 'EOF'
service:
  # For a Cassandra running outside a container
  - type: "cassandra"
    address: "127.0.0.1"
    port: 9042
    jmx_port: 7199
    jmx_username: "cassandra"  # by default, no authentication is done
    jmx_password: "cassandra"

  # For an additional Cassandra running outside a container
  - type: "cassandra"
    instance: "NAME_ASSOCIATED_WITH_YOUR_ADDITIONAL_SERVICE"
    address: "127.0.0.1"
    port: 9043
    jmx_port: 7200
    jmx_username: "cassandra"  # by default, no authentication is done
    jmx_password: "cassandra"

  # For a Cassandra running in a Docker container
  - type: "cassandra"
    instance: "CONTAINER_NAME"
    address: "172.17.0.2"
    port: 9042
    jmx_port: 7199
    jmx_username: "cassandra"  # by default, no authentication is done
    jmx_password: "cassandra"
EOF

Windows

Create or edit C:\ProgramData\glouton\conf.d\99-cassandra.conf:

service:
  # For a Cassandra running outside a container
  - type: "cassandra"
    address: "127.0.0.1"
    port: 9042
    jmx_port: 7199
    jmx_username: "cassandra"  # by default, no authentication is done
    jmx_password: "cassandra"

  # For an additional Cassandra running outside a container
  - type: "cassandra"
    instance: "NAME_ASSOCIATED_WITH_YOUR_ADDITIONAL_SERVICE"
    address: "127.0.0.1"
    port: 9043
    jmx_port: 7200
    jmx_username: "cassandra"  # by default, no authentication is done
    jmx_password: "cassandra"

  # For a Cassandra running in a Docker container
  - type: "cassandra"
    instance: "CONTAINER_NAME"
    address: "172.17.0.2"
    port: 9042
    jmx_port: 7199
    jmx_username: "cassandra"  # by default, no authentication is done
    jmx_password: "cassandra"

Glouton automatically detects configuration changes.

Built-in Metrics

Metric	Description
service_status	Status of Cassandra
cassandra_bloom_filter_false_ratio	Bloom filter false positive ratio in percent
cassandra_jvm_gc	Number of garbage collections per second
cassandra_jvm_gc_utilization	Garbage collection utilization in percent
cassandra_jvm_heap_used	Heap memory used in bytes
cassandra_jvm_non_heap_used	Non-heap memory used in bytes
cassandra_read_requests	Number of read requests per second
cassandra_read_time	Average time of read requests in seconds
cassandra_sstable	Number of SSTables
cassandra_write_requests	Number of write requests per second
cassandra_write_time	Average time of write requests in seconds

Per-Table Metrics

Bleemeo supports detailed monitoring of specific Cassandra tables. To enable this, add detailed_items to your service configuration.

Linux

sudo tee /etc/glouton/conf.d/99-cassandra.conf > /dev/null << 'EOF'
service:
  # For a Cassandra running outside a container
  - type: "cassandra"
    detailed_items:
      - "keyspace1.table1"
      - "keyspace2.table2"

  # For a Cassandra running in a Docker container
  - type: "cassandra"
    instance: "CONTAINER_NAME"
    detailed_items:
      - "keyspace1.table1"
      - "keyspace2.table2"
EOF

Windows

Create or edit C:\ProgramData\glouton\conf.d\99-cassandra.conf:

service:
  # For a Cassandra running outside a container
  - type: "cassandra"
    detailed_items:
      - "keyspace1.table1"
      - "keyspace2.table2"

  # For a Cassandra running in a Docker container
  - type: "cassandra"
    instance: "CONTAINER_NAME"
    detailed_items:
      - "keyspace1.table1"
      - "keyspace2.table2"

The following per-table metrics are gathered:

Metric	Description
cassandra_bloom_filter_false_ratio	Bloom filter false positive ratio in percent
cassandra_read_requests	Number of read requests per second
cassandra_read_time	Average time of read requests in seconds
cassandra_sstable	Number of SSTables
cassandra_write_requests	Number of write requests per second
cassandra_write_time	Average time of write requests in seconds

Monitoring Troubleshooting

See Troubleshoot a Service Check or Missing Metrics for what applies to every service: finding the address and port Glouton really uses, what each check message means, and how to read the collection error — which does not appear in the agent logs at the default level.

The Service Check is not OK

The check only opens the TCP port — it sends no command — so it fails when the port is wrong (Connection refused), when a firewall drops the packets (connection timed out after 10 seconds), or when the configured address cannot be parsed (Invalid TCP address). A green check proves the port is open, and nothing more.

Metrics are Missing

A passing check says nothing about the metrics: they are collected separately.

These metrics travel through JMX, which is a separate pipeline: Glouton writes /var/lib/jmxtrans/glouton-generated.json, jmxtrans connects to the JVM’s JMX port, and pushes the values back to Glouton. Every link has to be in place.

Check that the port is open and answers, from the machine where the agent runs. This only shows a JVM running on the host — for a containerized JVM, compare the address in discovery.txt with the ports the container itself listens on:

sudo ss -lntp | grep java

What you see	Cause	Fix
the bleemeo-agent-jmx package is missing	jmxtrans is not installed	sudo apt-get install bleemeo-agent-jmx (or yum) — see Java Monitoring with JMX
discovery.txt shows no JMX port for the service	No JMX port	Glouton reads it from the JVM command line — -Dcom.sun.management.jmxremote.port or -Dcassandra.jmx.remote.port. If neither is there, set jmx_port in the service configuration
connection refused by the JVM	JMX authentication is enabled	Set jmx_username and jmx_password
jmxtrans connects, then times out	JMX answered with an RMI address jmxtrans cannot reach	The JVM must advertise the address discovery.txt shows for the service, with -Djava.rmi.server.hostname
nothing at all, for a JVM in a container	jmxtrans connects to the container address, not to a port published on the host	Publishing the port does not help. The JMX and RMI ports have to answer on the container address, and -Djava.rmi.server.hostname has to be that address
the JVM has -Dcassandra.jmx.local.port=7199, which binds to loopback	Cassandra uses its local-only JMX port	Glouton auto-detects that option only when it reaches Cassandra on 127.0.0.1; otherwise switch Cassandra to -Dcassandra.jmx.remote.port

Glouton can only detect an option that actually reaches the JVM command line. Setting -Dcassandra.jmx.remote.port in SERVER_JVM_OPTS is not enough: it has to go in JVM_EXTRA_OPTS, which is a common reason for a Cassandra that stays without metrics while the option looks correctly set.

Per-table metrics are gathered only for the tables listed in detailed_items, each entry written as keyspace.table. Without them, only the cluster-wide values are collected — which is why a working JMX setup can still show fewer metrics than expected.