IAM statistics

IAM server has the ability of collecting its stats data of CPU, memory, requests, response and so on. IAM collects the stats data through StatsD Protocol . User can use these data to monitor the healthy of the IAM, and monitor the requests distribution.

Configuration settings

IAM has below statistics config settings, and all these settings have the statsConfig. prefix.

Setting Name	Default Value	Description
isEnabled	false	Enable the statistics feature or not
host	localhost	The host to send the stats to
port	8125	The port to send the stats to
protocol	udp	The protocol to be used by IAM to transfer the data to the statsd server. The value can be udp or tcp
interval	10000	The time interval in ms to flush the data to the host
format	telegraf	The stats data concrete format to be sent to the host. The value can be telgraf or statsd
debug	false	Output some debug info into console for debug the statistics issue

Enable the statistics

The statistics feature is disabled by default. User need do below steps to enable the statistics feature:

• Open IAM's local.properties from <iam_root>/config/ directory
• Add below 2 lints into it:

  statsConfig.isEnabled = true
  statsConfig.host = <Your_statsd_server_host>
  

Data collected by IAM

IAM collected 3 types metrics data.

• Counting Data
  • completed requests count
  • successed requests count
  • failed requestes count
• Timing Data
  • all requests time costs
  • ldap pool acquire and using time costs
• Gauges Data
  • CPU usages info
  • Memory usages info
  • Cache info
  • ldap pool info
  • request count in queue

Raw sample data

Here's an sample raw data sent from IAM to stats collector server:

• Counting data:

  request_completed,component=iam,host=test_server_local,process=master:6|c
  response_api_authorization_302,component=iam,host=test_server_local,process=master:1|c
  response_api_interaction_200,component=iam,host=test_server_local,process=master:1|c
  response_api_interaction_302,component=iam,host=test_server_local,process=master:1|c
  response_api_resume_302,component=iam,host=test_server_local,process=master:1|c
  response_api_token_200,component=iam,host=test_server_local,process=master:1|c
  response_api_introspection_200,component=iam,host=test_server_local,process=master:1|c
  

• Timing data:

  api_authorization,component=iam,host=test_server_local,process=master:49|ms
  api_interaction,component=iam,host=test_server_local,process=master:17|ms
  ldap_pool_acquire,component=iam,host=test_server_local,process=master:13|ms
  ldap_pool_use,component=iam,host=test_server_local,process=master:15|ms
  api_interaction,component=iam,host=test_server_local,process=master:95|ms
  api_resume,component=iam,host=test_server_local,process=master:45|ms
  api_token,component=iam,host=test_server_local,process=master:80|ms
  api_introspection,component=iam,host=test_server_local,process=master:37|ms
  

• Gauges data:

  request_in_queue,component=iam,host=test_server_local,process=master:0|g
  cache_count_UNID,component=iam,host=test_server_local,process=master:11|g
  cache_count_Session,component=iam,host=test_server_local,process=master:1|g
  cpu_user,component=iam,host=test_server_local,process=master:3.3440422741226064|g
  cpu_system,component=iam,host=test_server_local,process=master:0.4070695934939585|g
  memory_heapTotal,component=iam,host=test_server_local,process=master:39608320|g
  memory_heapUsed,component=iam,host=test_server_local,process=master:32721216|g
  ldap_user_pool_size,component=iam,host=test_server_local,process=master:1|g
  ldap_user_pool_available,component=iam,host=test_server_local,process=master:1|g
  ldap_admin_pool_size,component=iam,host=test_server_local,process=master:1|g
  ldap_admin_pool_available,component=iam,host=test_server_local,process=master:1|g
  

Next

Any Monitor system (Telegraf, Graphite, DogData, NewRelic, etc.) that supports "StatsD | DogStatsD" format input can collect these data and show them through dashboard.

As a reference, IAM-Dashboard project shows how to use Telegraf to collect data to InfluxDB and then use Grafana to present.