itm_exporter

command module
v0.0.0-...-8e9c542 Latest Latest
Warning

This package is not in the latest version of its module.

Go to latest
Published: Jan 3, 2021 License: MIT Imports: 23 Imported by: 0

README

ITM Exporter

ITM Exporter is a Prometheus exporter for IBM Tivoli Monitoring, IBM Application Performance Management (on-prem only) and IBM OMEGAMON. The exporter uses the ITM REST API in order to collect metrics from IBM ITM/APM/OMEGAMON. Note that ITM REST API is not officially supported.

How to use

Download and unpack the latest ITM Exporter release. Copy the exporter binary and config file config.yaml to your ITM TEPS server or IBM APM server. Edit the config file as described in the ITM exporter configuration section.

itm_exporter export

The above should start an exporter on default port 8000. You can check if it works using:

curl http://localhost:8000/metrics

Example output

# HELP itm_klzcpu_busycpu Busy CPU (Percent)
# TYPE itm_klzcpu_busycpu gauge
itm_klzcpu_busycpu{cpuid="0",originnode="rs41:LZ"} 48.19
itm_klzcpu_busycpu{cpuid="1",originnode="rs41:LZ"} 46.7
itm_klzcpu_busycpu{cpuid="2",originnode="rs41:LZ"} 44.33
itm_klzcpu_busycpu{cpuid="3",originnode="rs41:LZ"} 48.55
itm_klzcpu_busycpu{cpuid="Aggregate",originnode="rs41:LZ"} 46.94
# HELP itm_klzvm_memusedpct Memory Used (Percent)
# TYPE itm_klzvm_memusedpct gauge
itm_klzvm_memusedpct{originnode="rs41:LZ"} 99
# HELP itm_klzvm_vsfreepct Free Virtual Storage (Percent)
# TYPE itm_klzvm_vsfreepct gauge
itm_klzvm_vsfreepct{originnode="rs41:LZ"} 51
# HELP itm_klznet_transbps Bytes Transmitted Per Second
# TYPE itm_klznet_transbps gauge
itm_klznet_transbps{fname="eth0",originnode="rs41:LZ"} 0
itm_klznet_transbps{fname="eth1",originnode="rs41:LZ"} 545
itm_klznet_transbps{fname="lo",originnode="rs41:LZ"} 2.819566e+06
itm_klznet_transbps{fname="virbr0",originnode="rs41:LZ"} 0
itm_klznet_transbps{fname="virbr0-nic",originnode="rs41:LZ"} 0
# HELP itm_scrape_duration_seconds ITM attribute group scrape duration.
# TYPE itm_scrape_duration_seconds gauge
itm_scrape_duration_seconds{group="KLZCPU"} 1.987053388
itm_scrape_duration_seconds{group="KLZVM"} 1.231467763
itm_scrape_duration_seconds{group="KLZNET"} 1.172930966
# HELP itm_scrape_duration_seconds_all Total ITM scrape duration.
# TYPE itm_scrape_duration_seconds_all gauge
itm_scrape_duration_seconds_all 1.987057969
# HELP itm_scrape_status ITM attribute group scrape status.
# TYPE itm_scrape_status gauge
itm_scrape_status{group="KLZCPU"} 1
itm_scrape_status{group="KLZVM"} 1
itm_scrape_status{group="KLZNET"} 1

Check the all available CLI options with itm_exporter --help or itm_exporter --help-long.

Prometheus configuration

Add the following job to the scrape_configs section:

scrape_configs:
  - job_name: 'itm-exporter'
    scrape_interval: 60s
    scrape_timeout: 45s
    static_configs:
      - targets: ['<exporter_ip>:8000']

It is not recommended to specify scrape_interval less than 60s.

ITM Exporter configuration

Example config.yaml:

itm_server_url: "http://localhost:15200"
itm_server_user: "sysadmin"
itm_server_password: "pass"
connection_timeout: 8
collection_timeout: 40
groups: 
- name: "KLZCPU"
  datasets_uri: "/providers/itm.TEMS/datasources/TMSAgent.%25IBM.STATIC134/datasets"
  labels: ["CPUID", "ORIGINNODE"]
  metrics: ["BUSYCPU", "IDLECPU", "SYSCPU", "USRCPU", "WAITCPU"]
  managed_system_group: "*LINUX_SYSTEM"
- name: "KLZVM"
  datasets_uri: "/providers/itm.TEMS/datasources/TMSAgent.%25IBM.STATIC134/datasets"
  labels: ["ORIGINNODE"]
  metrics: ["MEMUSEDPCT", "VSFREEPCT"]
  managed_system_group: "*LINUX_SYSTEM"
- name: "KLZDISK"
  datasets_uri: '/providers/itm.TEMS/datasources/TMSAgent.%25IBM.STATIC134/datasets'
  labels: ["ORIGINNODE", "DSKNAME", "MOUNTPT"]
  metrics: ["DSKFREEPCT", "DSKUSEDPCT", "DSKFREE", "DSKUSED", "INDFREEPCT"]
  managed_system_group: "*LINUX_SYSTEM"
- name: "KLZNET"
  datasets_uri: '/providers/itm.TEMS_1%3A2-3/datasources/TMSAgent.%25IBM.STATIC134/datasets'
  labels: ["ORIGINNODE", "FNAME"]
  metrics: ["TRANSBPS"]
  managed_system_group: "*LINUX_SYSTEM"
- name: "msys"
  datasets_uri: "/providers/itm.TEMS_1%3A2-3/datasources/TMSAgent.%26IBM.STATIC000/datasets"
  labels: ["ORIGINNODE", "PRODUCT", "AFFPRODUCT", "VERSION", "OSPLATFORM", "NETADDR", "HOSTNAME"]
  metrics: ["AVAILABLE"]
  managed_system_group: "*TEMS"
  • itm_server_url - HTTP URL of your TEPS or APM Server, ex.: "http://localhost:15210"
  • itm_server_user - for example sysadmin for ITM v6 or smadmin for APM v8
  • itm_server_password
  • connection_timeout - maximum time allowed for a simple http request from itm_exporter to ITM CURI API.
  • collection_timeout - maximum time allowed for collecting the latest snapshot of metric values for a single Attribute Group.

The section groups: specifies which ITM/APM metrics should be collected and exposed by the exporter. ITM exporter concurrently collects metrics for every group.

  • datasets_uri - it is a part of the API request URL that identifies particular agent type. The exporter helps a bit in the identification of proper datasets_uri for the agent type you'd like to collect. Run the following command:
  itm_exporter listAgentTypes --temsName=TEMS 

where temsName is your ITM TEMS label like TEMS or KD8 if you connect to APM v8 server. Example output:

+---------------------------------+--------------------------------------------------------------------+
|           AGENT TYPE            |                            DATASET URI                             |
+---------------------------------+--------------------------------------------------------------------+
| Tivoli Enterprise Portal Server | /providers/itm.TEMS/datasources/TMSAgent.%25IBM.STATIC153/datasets |
| Windows OS                      | /providers/itm.TEMS/datasources/TMSAgent.%25IBM.STATIC021/datasets |
| All Managed Systems             | /providers/itm.TEMS/datasources/TMSAgent.%26IBM.STATIC000/datasets |
| Summarization and Pruning Agent | /providers/itm.TEMS/datasources/TMSAgent.%25IBM.STATIC066/datasets |
| Warehouse Proxy                 | /providers/itm.TEMS/datasources/TMSAgent.%25IBM.STATIC122/datasets |
| Linux OS                        | /providers/itm.TEMS/datasources/TMSAgent.%25IBM.STATIC134/datasets |
| UISolution.manager              | /providers/itm.TEMS/datasources/UISolution.manager/datasets        |
+---------------------------------+--------------------------------------------------------------------+
  • name - name of the group you'd like to collect. You can list attribute group names with the following command (example for Linux OS dataset):
itm_exporter listAttributeGroups --dataset=/providers/itm.TEMS/datasources/TMSAgent.%25IBM.STATIC134/datasets

Example output:

+--------------------------------------+-----------------+
|             DESCRIPTION              | ATTRIBUTE GROUP |
+--------------------------------------+-----------------+
| Linux Network                        | KLZNET          |
| Agent Operations Log                 | OPLOG           |
| Linux NFS Statistics (Superseded)    | LNXNFS          |
| Linux Sockets Detail                 | KLZSOCKD        |
| Linux Disk Usage Trends (Superseded) | LNXDU           |
| Linux Group                          | LNXGROUP        |
| Linux Process (Superseded)           | LNXPROC         |
| Linux CPU                            | KLZCPU          |
| Linux IP Address                     | LNXIPADDR       |
| Linux Process                        | KLZPROC         |
| Linux File Comparison                | LNXFILCMP       |
| Linux Sockets Status                 | KLZSOCKS        |
| Linux System Statistics (Superseded) | LNXSYS          |
| Situation Event Information          | events          |
| Linux IO Ext                         | KLZIOEXT        |
| Linux NFS Statistics                 | KLZNFS          |
| Linux CPU Averages (Superseded)      | LNXCPUAVG       |
| CustomScriptsRuntime Sampled         | KLZSCRTSM       |
| Linux OS Config                      | LNXOSCON        |
| Linux Host Availability              | LNXPING         |
| Linux Sockets Detail (Superseded)    | LNXSOCKD        |
| Configuration Information            | KLZPASCAP       |
| Linux VM Stats (Superseded)          | LNXVM           |
| Linux RPC Statistics (Superseded)    | LNXRPC          |
| Linux System Statistics              | KLZSYS          |
| Linux File Information               | LNXFILE         |
| Linux Swap Rate                      | KLZSWPRT        |
| Alerts Table                         | KLZPASALRT      |
| Linux Sockets Status (Superseded)    | LNXSOCKS        |
| Managed System Information           | msys            |
| CustomScriptsRuntime                 | KLZSCRRTM       |
| Linux IO Ext (Superseded)            | LNXIOEXT        |
| Linux TCP Statistics                 | KLZTCP          |
| CustomScripts                        | KLZSCRPTS       |
| Linux Disk IO                        | KLZDSKIO        |
| Agent Availability Management Status | KLZPASMGMT      |
| Linux All Users                      | LNXALLUSR       |
| Linux RPC Statistics                 | KLZRPC          |
| Linux Disk Usage Trends              | KLZDU           |
| Linux User Login                     | KLZLOGIN        |
| Linux CPU Config                     | LNXCPUCON       |
| Linux Disk (Superseded)              | LNXDISK         |
| Linux Machine Information            | LNXMACHIN       |
| Linux Swap Rate (Superseded)         | LNXSWPRT        |
| Linux Disk                           | KLZDISK         |
| Managed System Groups                | mgrp            |
| Linux Process User Info (Superseded) | LNXPUSR         |
| Agent Active Runtime Status          | KLZPASSTAT      |
| Linux Process User Info              | KLZPUSR         |
| Linux CPU Averages                   | KLZCPUAVG       |
| Linux Network (Superseded)           | LNXNET          |
| Linux Disk IO (Superseded)           | LNXDSKIO        |
| Linux LPAR                           | KLZLPAR         |
| Linux VM Stats                       | KLZVM           |
| Linux CPU (Superseded)               | LNXCPU          |
| Linux File Pattern                   | LNXFILPAT       |
| Linux User Login (Superseded)        | LNXLOGIN        |
| Situation Advice                     | advice          |
+--------------------------------------+-----------------+
  • labels - list of attributes that should be mapped as Prometheus metric labels (typically string attributes that identify source of the metric like ORIGINNODE, CPUID or MOUNTPT). As a rule of thumb you should always use at least ORIGINNODE and all other metrics that are primary keys within the attribute group as labels (listAttributes command shows which metrics are primary keys), otherwise you may see error about duplicate metrics in the /metrics output.
  • metrics - numeric metrics names you'd like to collect.

Attribute names (for mapping with both labels and metrics) can be listed with the following command (example for KLZCPU attribute group within Linux OS dataset):

itm_exporter listAttributes --attributeGroup=KLZCPU --dataset=/providers/itm.TEMS/datasources/TMSAgent.%25IBM.STATIC134/datasets

Example output:

+------------------------------+------------+-------------+
|         DESCRIPTION          | ATTRIBUTES | PRIMARY KEY |
+------------------------------+------------+-------------+
| System Name                  | ORIGINNODE | false       |
| Time Stamp                   | TIMESTAMP  | false       |
| CPU ID                       | CPUID      | true        |
| User CPU (Percent)           | USRCPU     | false       |
| User Nice CPU (Percent)      | USRNCPU    | false       |
| System CPU (Percent)         | SYSCPU     | false       |
| Idle CPU (Percent)           | IDLECPU    | false       |
| Busy CPU (Percent)           | BUSYCPU    | false       |
| I/O Wait (Percent)           | WAITCPU    | false       |
| User to System CPU (Percent) | USRSYSCPU  | false       |
| Steal CPU (Percent)          | STEALCPU   | false       |
| Recording Time               | WRITETIME  | false       |
+------------------------------+------------+-------------+
  • managed_system_group - the name of the managed system group, grouping agents in scope of the collection.

ITM Exporter CLI options

usage: itm_exporter [<flags>] <command> [<args> ...]

ITM exporter for Prometheus.

Flags:
      --help        Show context-sensitive help (also try --help-long and --help-man).
  -c, --configFile="config.yaml"
                    Configuration file
      --web.listen-address=":8000"
                    The address to listen on for HTTP requests.
  -v, --verboseLog  Verbose logging for export and diagnostic modes.

Commands:
  help [<command>...]
    Show help.


  listAttributes --attributeGroup=ATTRIBUTEGROUP --dataset=DATASET
    List available attributes for the given attribute group.

    -g, --attributeGroup=ATTRIBUTEGROUP
                           Attribute group
    -d, --dataset=DATASET  Dataset (Agent type) URI. You can find it using command: 'itm_exporter listAgentTypes'. Example Dataset URI for Linux OS Agent:
                           '/providers/itm.TEMS/datasources/TMSAgent.%25IBM.STATIC134/datasets'.

  listAttributeGroups --dataset=DATASET [<flags>]
    List available Attribute Groups for the given dataset.

    -d, --dataset=DATASET  Dataset (Agent type) URI. You can find it using command: 'itm_exporter listAgentTypes'. Example Dataset URI for Linux OS Agent:
                           '/providers/itm.TEMS/datasources/TMSAgent.%25IBM.STATIC134/datasets'.
    -l, --long             List Attributes for every Attribute Group in dataset

  listAgentTypes --temsName=TEMSNAME
    Lists datasets (agent types).

    -t, --temsName=TEMSNAME  ITM TEMS label (specify KD8 for APMv8).

  export
    Start itm_exporter in exporter mode.


  test --file=FILE
    Start itm_exporter in diagnostic mode.

    --file=FILE  JSON response

Prometheus Quick Start

If you are not familiar with Prometheus, a good option is to start with full Prometheus/Grafana stack running in Docker Compose.

  1. Install Docker and Docker Compose: https://docs.docker.com/compose/install/
  2. git clone https://github.com/vegasbrianc/prometheus
  3. cd prometheus
  4. vi prometheus/prometheus.yml and add itm-exporter job as describe above.
  5. docker-compose up -d
  6. Check Prometheus URL via web browser: http://localhost:9090/targets ad make sure that Prometheus server can scrape itm-exporter
  7. If the job status is UP, access Grafana via web browser: http://localhost:9090/ (admin/foobar).
  8. Import both dashboards included in this repo.

License

Under MIT.

Documentation

The Go Gopher

There is no documentation for this package.

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL