check_couchbase.yaml

#
#
# COUCHBASE CONNECTION
#
#

# IP address or hostname of the Couchbase node to monitor
couchbase_host: localhost

# Admin username
#  < Couchbase 5: recommended to use the Read-Only Administrator
#  >= Couchbase 5: admin user with the the Data Monitoring role
couchbase_user: readonly

# Admin password
couchbase_password: secret

# Use SSL when talking to Couchbase
# Must also use the correct port above
couchbase_ssl: False

# Override the default Couchbase ports
#couchbase_admin_port: 8091
#couchbase_admin_port_ssl: 18091

#couchbase_query_port: 8093
#couchbase_query_port_ssl: 18093

#couchbase_fts_port: 8094
#couchbase_fts_port: 18094


#
#
# MONITOR CONNECTION
#
#

# The type of monitoring system in use
#  Options are:
#   nagios
#   graphite
monitor_type: nagios

# IP address or hostname of the monitor server
monitor_host: localhost

# Monitor service port
#  If Nagios, defaults are:
#   nsca: 5667
#   nsca-ng: 5668
#  If Graphite, default is:
#    carbon: 2003
monitor_port: 5668


#
#
# NAGIOS CONFIGURATION 
#
#

# Path to the send_nsca binary
# Note that the user running this script must be able to read /etc/send_nsca.cfg
nagios_nsca_path: /usr/sbin/send_nsca


#
#
# SERVICE NAME
#
#
#  These settings define how the service name should be constructed.
#  The service name format will be:
#   {prefix}.{cluster name}.{label} - {metric description}

# Service prefix
service_prefix: Couchbase

# Include the cluster name in the service description
service_include_cluster_name: True

# Include the label in the service description
#  Data service label is the bucket name
#  Other services use the service name, e.g. "query" or "xdcr"
service_include_label: True


#
#
# METRICS AND THRESHOLDS
#
#
#  Metrics are monitored by service: data, query, index, fts, xdcr, node
#
#  Metrics have 5 possible values:
#  Required:
#    metric: the Couchbase REST API metric to check
#    description: used to construct the service name
#  Optional:
#    warn: the warning alert threshold
#    crit: the critical alert threshold
#    op: operator used to compare the current value against the thresholds
#
#  The statuses sent to the monitor are defined by the configured thresholds along
#  with an implicit "OK".  Not specifying thresholds will result in "OK"
#  always being sent to the monitor.
#
#  "op" allows you to define the operator used when evaluating thresholds
#  againt the current value.  Options are: ">", ">=", "=", "<=", "<"
#  The default operator is ">="

# Data service
#  Metrics are configurable by bucket, allowing thresholds to be customized.
#
#  The bucket name is specified by the "bucket" parameter.
#
#  Special name "_all" will evaluate the configured metrics and thresholds for
#  all buckets
#
#  Metric names are documented here:
#  https://github.com/couchbase/ep-engine/blob/master/docs/stats.org
#
#  The following calculated metrics have been added:
#    percent_quota_utilization: mem_used / ep_mem_high_wat
#    percent_metadata_utilization: ep_meta_data_memory / ep_mem_high_wat
#    disk_write_queue: ep_queue_size + ep_flusher_todo
#    total_ops: cmd_get + cmd_set + incr_misses + incr_hits + decr_misses + decr_hits + delete_misses + delete_hits
data:
  - bucket: _all
    metrics:
    - metric: percent_quota_utilization
      description: percent bucket quota used
      warn: 80
      crit: 90 
    - metric: percent_metadata_utilization
      description: percent bucket quota used by metadata
      warn: 10
      crit: 20
    - metric: disk_write_queue
      description: items in disk write queue
      warn: 10000
      crit: 50000
    - metric: total_ops
      description: total ops per second
      warn: 10000
      crit: 20000
    - metric: cmd_get
      description: gets per second
      warn: 5000
      crit: 10000
    - metric: cmd_set
      description: sets per second
      warn: 5000
      crit: 10000
    - metric: delete_hits
      description: deletes per second
      warn: 500
      crit: 1000
    - metric: ep_cache_miss_rate
      description: cache miss ratio 
      warn: 1
      crit: 10
    - metric: couch_docs_fragmentation
      description: percent data fragmentation
      warn: 35
      crit: 50
    - metric: couch_views_fragmentation
      description: percent views fragmentation
      warn: 35
      crit: 50
    - metric: curr_connections
      description: client connections
      warn: 500
      crit: 1000  # Backoffs are sent at 10,0000 
    - metric: ep_dcp_replica_items_remaining
      description: items in internal replication queue
      warn: 2500
      crit: 5000
    - metric: ep_dcp_2i_items_remaining
      description: items in 2i indexer queue
      warn: 5000
      crit: 10000
    - metric: ep_dcp_views_items_remaining
      description: items in views indexer queue
      warn: 5000
      crit: 10000
    - metric: ep_dcp_replica_backoff
      description: internal replication backoffs
      crit: 1
    - metric: ep_dcp_xdcr_backoff
      description: XDCR backoffs
      crit: 1
    - metric: vb_avg_total_queue_age
      description: disk write queue average age
      warn: 5
      crit: 10
    - metric: ep_oom_errors
      description: out of memory errors
      crit: 1
    - metric: ep_tmp_oom_errors
      description: temporary out of memory errors
      crit: 1
    - metric: vb_active_resident_items_ratio
      description: percent active items in memory
      warn: 50
      crit: 15  # Should never be less than 15%
      op: "<"
    - metric: vb_replica_resident_items_ratio
      description: percent replica items in memory
      warn: 50
      crit: 15  # Should never be less than 15%
      op: "<"

# Query service
#  All metric names documented here (under Vitals):
#  https://developer.couchbase.com/documentation/server/current/tools/query-monitoring.html
query:
  - metric: request_timer.75%
    description: 75th percentile query response time
    warn: 100  # Milliseconds
    crit: 200  # Milliseconds
  - metric: request_timer.95%
    description: 95th percentile query response time
    warn: 200  # Milliseconds
    crit: 300  # Milliseconds
  - metric: request_timer.99%
    description: 99th percentile query response time
    warn: 400  # Milliseconds
    crit: 500  # Milliseconds
  - metric: active_requests.count
    description: active N1QL requests
    warn: 1000
    crit: 1500
  - metric: request_rate.1m.rate
    description: query throughput 1 minute
    warn: 900
    crit: 950
  - metric: request_rate.5m.rate
    description: query throughput 5 minute
    warn: 800
    crit: 850
  - metric: request_rate.15m.rate
    description: query throughput 15 minute
    warn: 700
    crit: 750

# Full Text Search (FTS) service
#  All metric names documented here (under Vitals):
#  https://developer.couchbase.com/documentation/server/current/rest-api/rest-fts-indexing.html#topic_hpd_2y4_1v__g-api-stats
#
#  Note that these metrics will be applied to each FTS index independently
fts:
  - metric: num_mutations_to_index
    description: items in FTS indexer queue
    warn: 2000
    crit: 5000
#  - metric: total_queries_slow  # commented until a rate metric is available
#    description: FTS slow searches  # queries that take longer than 5 seconds 
#    warn: 100
#    crit: 1000
#  - metric: total_queries_timeout
#    description: FTS search timeouts # commented until a rate metric is available
#    warn: 100
#    crit: 1000
#  - metric: total_queries_error # commented until a rate metric is available
#    description: FTS search errors 
#    warn: 100
#    crit: 1000

# XDCR
#  All metric names documented here:
#  https://developer.couchbase.com/documentation/server/current/rest-api/rest-xdcr-statistics.html
#
#  Note that these metrics will be applied to each XDCR replication independently
xdcr:
  - metric: status
    description: replication status
    warn: paused
    crit: notRunning
    op: "="
  - metric: changes_left
    description: documents pending replication
    warn: 2500
    crit: 5000
  - metric: bandwidth_usage
    description: bytes replicated per second
    warn: 12500000  # 100 Mbps
    crit: 25000000  # 200 Mbps

# Node Stats
node:
  - metric: status
    description: health status
    warn: warmup
    crit: unhealthy
    op: "="
  - metric: clusterMembership
    description: cluster membership
    warn: inactiveAdded
    crit: inactiveFailed
    op: "="

#
#
# PYTHON LOGGING
#
#
logging:
  version: 1
  formatters:
    simple:
      format: "%(asctime)s %(levelname)s %(message)s"
  handlers:
    console:
      class: logging.StreamHandler
      level: ERROR
      formatter: simple
      stream: ext://sys.stdout
  root:
    level: DEBUG
    handlers: [console]