Azure Managed Instance for Apache Cassandra Monitoring

From both a data and infrastructure perspective, this Prometheus Extension 2.0 allows you to monitors and analyze the activity of your Apache Cassandra clusters. It visualize your cluster's health and shows metrics like CPU, connectivity, request latency, suspension, and garbage collection time. Additionally, with Davis, it automatically detects performance problems and provides precise root cause analysis.

Prerequisites

Setup

  1. Create an Ubuntu virtual machine in the same virtual network as your Azure Managed Instance for Apache Cassandra.
  2. Ensure Docker is installed on your virtual machine.
  3. Create a file named prometheus.yml on your virtual machine with the contents below.
    Add every Cassandra Node IP address and port 9443 in the static_configs section. The IP addresses can be gathered from the Data Center section of the Azure Portal for your Cassandra Cluster. 
    static_configs:
        - targets: ["<Node_IP_1>:9443", "<Node_IP_2>:9443", "<Node_IP_N>:9443"]
global:
  scrape_interval: 15s
  scrape_timeout: 10s
  evaluation_interval: 15s


alerting:
  alertmanagers:
  - static_configs:
    - targets: []
    scheme: http
    timeout: 10s


scrape_configs:
  - job_name: prometheus
    scrape_interval: 15s
    scrape_timeout: 15s
    metrics_path: /metrics
    scheme: http
    static_configs:
    - targets:
      - localhost:9090


  - job_name: "mcac"
    scrape_interval: 15s
    scrape_timeout:  15s
    static_configs:
      - targets: ["<Node_IP_1>:9443", "<Node_IP_2>:9443", "<Node_IP_N>:9443"]
    honor_labels: true
    honor_timestamps: false
    scheme: https
    tls_config:
            insecure_skip_verify: true
    metric_relabel_configs:
    #drop metrics we can calculate from prometheus directly
    - source_labels: [__name__]
      regex: .*rate_(mean|1m|5m|15m)
      action: drop
    #save the original name for all metrics
    - source_labels: [__name__]
      regex: (collectd_mcac_.+)
      target_label: prom_name
      replacement: ${1}
    - source_labels: ["prom_name"]
      regex: .+_bucket_(\d+)
      target_label: le
      replacement: ${1}
    - source_labels: ["prom_name"]
      regex: .+_bucket_inf
      target_label: le
      replacement: +Inf
    - source_labels: ["prom_name"]
      regex: .*_histogram_p(\d+)
      target_label: quantile
      replacement: .${1}
    - source_labels: ["prom_name"]
      regex: .*_histogram_min
      target_label: quantile
      replacement: "0"
    - source_labels: ["prom_name"]
      regex: .*_histogram_max
      target_label: quantile
      replacement: "1"
    #Table Metrics *ALL* we can drop
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.table\.(\w+)
      action: drop
    #Table Metrics
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.table\.(\w+)\.(\w+)\.(\w+)
      target_label: table
      replacement: ${3}
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.table\.(\w+)\.(\w+)\.(\w+)
      target_label: keyspace
      replacement: ${2}
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.table\.(\w+)\.(\w+)\.(\w+)
      target_label: __name__
      replacement: mcac_table_${1}
    #Keyspace Metrics
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.keyspace\.(\w+)\.(\w+)
      target_label: keyspace
      replacement: ${2}
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.keyspace\.(\w+)\.(\w+)
      target_label: __name__
      replacement: mcac_keyspace_${1}
    #ThreadPool Metrics (one type is repair.task so we just ignore the second part)
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.thread_pools\.(\w+)\.(\w+)\.(\w+).*
      target_label: pool_type
      replacement: ${2}
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.thread_pools\.(\w+)\.(\w+)\.(\w+).*
      target_label: pool_name
      replacement: ${3}
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.thread_pools\.(\w+)\.(\w+)\.(\w+).*
      target_label: __name__
      replacement: mcac_thread_pools_${1}
    #ClientRequest Metrics
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.client_request\.(\w+)\.(\w+)$
      target_label: request_type
      replacement: ${2}
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.client_request\.(\w+)\.(\w+)$
      target_label: __name__
      replacement: mcac_client_request_${1}
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.client_request\.(\w+)\.(\w+)\.(\w+)$
      target_label: cl
      replacement: ${3}
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.client_request\.(\w+)\.(\w+)\.(\w+)$
      target_label: request_type
      replacement: ${2}
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.client_request\.(\w+)\.(\w+)\.(\w+)$
      target_label: __name__
      replacement: mcac_client_request_${1}_cl
    #Cache Metrics
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.cache\.(\w+)\.(\w+)
      target_label: cache_name
      replacement: ${2}
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.cache\.(\w+)\.(\w+)
      target_label: __name__
      replacement: mcac_cache_${1}
    #CQL Metrics
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.cql\.(\w+)
      target_label: __name__
      replacement: mcac_cql_${1}
    #Dropped Message Metrics
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.dropped_message\.(\w+)\.(\w+)
      target_label: message_type
      replacement: ${2}
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.dropped_message\.(\w+)\.(\w+)
      target_label: __name__
      replacement: mcac_dropped_message_${1}
    #Streaming Metrics
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.streaming\.(\w+)\.(.+)$
      target_label: peer_ip
      replacement: ${2}
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.streaming\.(\w+)\.(.+)$
      target_label: __name__
      replacement: mcac_streaming_${1}
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.streaming\.(\w+)$
      target_label: __name__
      replacement: mcac_streaming_${1}
    #CommitLog Metrics
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.commit_log\.(\w+)
      target_label: __name__
      replacement: mcac_commit_log_${1}
    #Compaction Metrics
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.compaction\.(\w+)
      target_label: __name__
      replacement: mcac_compaction_${1}
    #Storage Metrics
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.storage\.(\w+)
      target_label: __name__
      replacement: mcac_storage_${1}
    #Batch Metrics
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.batch\.(\w+)
      target_label: __name__
      replacement: mcac_batch_${1}
    #Client Metrics
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.client\.(\w+)
      target_label: __name__
      replacement: mcac_client_${1}
    #BufferPool Metrics
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.buffer_pool\.(\w+)
      target_label: __name__
      replacement: mcac_buffer_pool_${1}
    #Index Metrics
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.index\.(\w+)
      target_label: __name__
      replacement: mcac_sstable_index_${1}
    #HintService Metrics
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.hinted_hand_off_manager\.([^\-]+)-(\w+)
      target_label: peer_ip
      replacement: ${2}
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.hinted_hand_off_manager\.([^\-]+)-(\w+)
      target_label: __name__
      replacement: mcac_hints_${1}
    #HintService Metrics
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.hints_service\.hints_delays\-(\w+)
      target_label: peer_ip
      replacement: ${1}
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.hints_service\.hints_delays\-(\w+)
      target_label: __name__
      replacement: mcac_hints_hints_delays
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.hints_service\.([^\-]+)
      target_label: __name__
      replacement: mcac_hints_${1}
    # Misc
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.memtable_pool\.(\w+)
      target_label: __name__
      replacement: mcac_memtable_pool_${1}
    - source_labels: ["mcac"]
      regex: com\.datastax\.bdp\.type\.performance_objects\.name\.cql_slow_log\.metrics\.queries_latency
      target_label: __name__
      replacement: mcac_cql_slow_log_query_latency
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.read_coordination\.(.*)
      target_label: read_type
      replacement: $1
    - source_labels: ["mcac"]
      regex: org\.apache\.cassandra\.metrics\.read_coordination\.(.*)
      target_label: __name__
      replacement: mcac_read_coordination_requests
    #GC Metrics
    - source_labels: ["mcac"]
      regex: jvm\.gc\.(\w+)\.(\w+)
      target_label: collector_type
      replacement: ${1}
    - source_labels: ["mcac"]
      regex: jvm\.gc\.(\w+)\.(\w+)
      target_label: __name__
      replacement: mcac_jvm_gc_${2}
    #JVM Metrics
    - source_labels: ["mcac"]
      regex: jvm\.memory\.(\w+)\.(\w+)
      target_label: memory_type
      replacement: ${1}
    - source_labels: ["mcac"]
      regex: jvm\.memory\.(\w+)\.(\w+)
      target_label: __name__
      replacement: mcac_jvm_memory_${2}
    - source_labels: ["mcac"]
      regex: jvm\.memory\.pools\.(\w+)\.(\w+)
      target_label: pool_name
      replacement: ${2}
    - source_labels: ["mcac"]
      regex: jvm\.memory\.pools\.(\w+)\.(\w+)
      target_label: __name__
      replacement: mcac_jvm_memory_pool_${2}
    - source_labels: ["mcac"]
      regex: jvm\.fd\.usage
      target_label: __name__
      replacement: mcac_jvm_fd_usage
    - source_labels: ["mcac"]
      regex: jvm\.buffers\.(\w+)\.(\w+)
      target_label: buffer_type
      replacement: ${1}
    - source_labels: ["mcac"]
      regex: jvm\.buffers\.(\w+)\.(\w+)
      target_label: __name__
      replacement: mcac_jvm_buffer_${2}
    #Append the prom types back to formatted names
    - source_labels: [__name__, "prom_name"]
      regex: (mcac_.*);.*(_micros_bucket|_bucket|_micros_count_total|_count_total|_total|_micros_sum|_sum|_stddev).*
      separator: ;
      target_label: __name__
      replacement: ${1}${2}
    - regex: prom_name
      action: labeldrop

  1. Start your Prometheus server Docker container.

    Be sure to change the path in the command below to point to the prometheus.yml file from above.

    docker run \
        -d \
        -p 9090:9090 \
        -v /path/to/prometheus.yml:/etc/prometheus/prometheus.yml \
        prom/prometheus

  2. If your virtual machine is not available from the internet, install a Dynatrace Environment ActiveGate on your Ubuntu VM.
    Recommended: Set the group property on the installation.

Enable and configure extension

  1. In Dynatrace Hub, select Azure Managed Instance for Apache Cassandra.

  2. Enable the extension.

  3. Verify that the Prometheus endpoint publishes the Cassandra metrics. Use either of these queries:

    {__name__=~"mcac.*"}

    http://<Prometheus Server URL>:9090/api/v1/query?query=%7B__name__%3D%7E%22mcac.*%22%7D

  4. Add the endpoint of your Prometheus server to the Extension Monitoring Configuration:

    http://<Prometheus Server URL>:9090/api/v1

    The <Prometheus Server URL> does not need to be public. If you install your ActiveGate on the same VM or same VNet as the Prometheus server, localhost or a private IP can be used.

  5. Select the ActiveGate group on which to enable this extension.

  6. Add a Monitoring Configuration description and select the Feature Sets of the metrics you'd like to collect.

  7. A dashboard named Azure Managed Instance for Apache Cassandra Overview is provided with the extension.

Metrics

Available metrics are listed below.

Cluster node metrics

Metric Name
Metric Key
Description
Storage Load
com.dynatrace.extension.prometheus.azure_cassandra_storage_load
Size, in bytes, of the on-disk data size this node manages.
Storage Exceptions
com.dynatrace.extension.prometheus.azure_cassandra_storage_exceptions.count
Number of internal exceptions caught. In normal operation, this should be zero.
Commit Log Pending Tasks
com.dynatrace.extension.prometheus.azure_cassandra_commit_log_pending_tasks
Number of commit log messages written but yet to be fsync'd.
Commit Log Completed Tasks Total
com.dynatrace.extension.prometheus.azure_cassandra_commit_log_completed_tasks_total.count
Total number of commit log messages written since start/restart.
Buffer Pool Size
com.dynatrace.extension.prometheus.azure_cassandra_buffer_pool_size
Size, in bytes, of the managed buffer pool.
Buffer Pool Misses Total
com.dynatrace.extension.prometheus.azure_cassandra_buffer_pool_misses_total.count
The number of misses in the pool. The higher this is, the more allocations incurred.

Keyspace metrics

Metric Name
Metric Key
Description
Keyspace All Memtables Live Data Size
com.dynatrace.extension.prometheus.azure_cassandra_keyspace_all_memtables_live_data_size
Total amount of live data stored in the memtables (2i and pending flush memtables included) that resides off-heap, excluding any data structure overhead.
Keyspace Bloom Filter Disk Space Used
com.dynatrace.extension.prometheus.azure_cassandra_keyspace_bloom_filter_disk_space_used
Disk space used by bloom filter (in bytes).
Keyspace Live Disk Space Used
com.dynatrace.extension.prometheus.azure_cassandra_keyspace_live_disk_space_used
Disk space used by SSTables belonging to this table (in bytes).
Keyspace Memtable Columns Count
com.dynatrace.extension.prometheus.azure_cassandra_keyspace_memtable_columns_count.gauge
Total number of columns present in the memtable.
Keyspace Memtable Live Data Size
com.dynatrace.extension.prometheus.azure_cassandra_keyspace_memtable_live_data_size
Total amount of live data stored in the memtable, excluding any data structure overhead.
Keyspace Memtable Switch Count
com.dynatrace.extension.prometheus.azure_cassandra_keyspace_memtable_switch_count.gauge
Number of times that flush has resulted in the memtable being switched out.

Table metrics

Metric Name
Metric Key
Description
Table Bloom Filter Disk Space Used
com.dynatrace.extension.prometheus.azure_cassandra_table_bloom_filter_disk_space_used
Disk space used by bloom filter (in bytes).
Table Bloom Filter False Positives
com.dynatrace.extension.prometheus.azure_cassandra_table_bloom_filter_false_positives
Number of false positives on table's bloom filter.
Table Bloom Filter False Ratio
func:com.dynatrace.extension.prometheus.azure_cassandra_table_bloom_filter_false_ratio
False positive ratio of table's bloom filter.
Table Bytes Flushed Total
com.dynatrace.extension.prometheus.azure_cassandra_table_bytes_flushed_total.count
Total number of bytes flushed since server start/restart.
Table Compaction Bytes Written Total
com.dynatrace.extension.prometheus.azure_cassandra_table_compaction_bytes_written_total.count
Total number of bytes compacted since server start/restart.
Table Compression Ratio
func:com.dynatrace.extension.prometheus.azure_cassandra_table_compression_ratio
Current compression ratio for all SSTables.