Skip to content

Welcome to CMT Documentation

Current release : 2.3 - 2022/01/23

Overwiew - CMT is a simple monitoring agent

CMT is a small footprint agent, with a single file configuration, available for most standard Linux OS. Once deployed on a system, it collects various data and monitors local/remote services.

Send to ElasticSearch/InfluxDB/Graylog

Collected data is sent to Metrology servers like ElasticSearch, InfluxDB V1/V2 or Graylog. These data points may carry alert information or warning levels for central alerting or display (Kibana for example).

Alerts and Pager to Teams and PagerDuty

If various conditions are met (or not met), CMT can also send direct alerts / pagers notifications to Teams channel.

Use as CLI

For sysadmin, CMT can be used as command line tool on each server during various sysadmin or software configuration tasks. One single run provides an immediate health check of all the configured check items.

$ cmt -s

CMT - (c) - Version 2.3 - 2022/01/23
cmt_group      :  cavaliba
cmt_node       :  vmxu3
config file    :  /opt/cmt/conf.yml

OK       boottime     boot : 0 days since last reboot - 4:22:08 sec.
OK       load         load 1/5/15 min : 0.19  0.12  0.1 (2 cpus)
OK       certificate  56 day(s) left for SSL certificate on 
OK       certificate  56 day(s) left for SSL certificate on 
OK       certificate  307 day(s) left for SSL certificate on 
CRITICAL certificate  no ssl connection to
OK       certificate  143 day(s) left for SSL certificate on 
OK       cpu          cpu usage : 2.0 %
OK       disk         disk / - used: 44.6 % - used: 28.5 GB - free: 35.4 GB - total: 67.4 GB 
OK       disk         disk /boot - used: 44.6 % - used: 28.5 GB - free: 35.4 GB - total: 67.4 GB 
OK       folder       test_permission /opt/cmt/testdata/permission.txt OK - 1 files, 0 dirs, 0 bytes - targets 1/1
CRITICAL folder       folder /opt/cmt/testdata/permission.txt : incorrect permission for /opt/cmt/testdata/permission.txt: found -rw-rw-r-- , expected -rw-rw-rw-
OK       folder       test_permissions /opt/cmt/testdata/permissions OK - 4 files, 0 dirs, 16 bytes - targets 3/3
OK       folder       test_extension /opt/cmt/testdata OK - 2 files, 6 dirs, 0 bytes - targets 0/0
OK       folder       test_regexp /opt/cmt/testdata OK - 2 files, 6 dirs, 0 bytes - targets 0/0
OK       folder       test_regexp_ext /opt/cmt/testdata OK - 1 files, 6 dirs, 0 bytes - targets 0/0
WARNING  folder       test_wrong_target /opt/cmt/testdata : unknown target is_blabla
OK       folder       test_hasfile /opt/cmt/testdata OK - 7 files, 6 dirs, 11004 bytes - targets 1/1
OK       folder       test_age_min /opt/cmt/testdata OK - 7 files, 6 dirs, 11004 bytes - targets 1/1
CRITICAL folder       test_age_max /opt/cmt/testdata : some files are too old (25324485 sec)
OK       folder       test_files_min /opt/cmt/testdata OK - 7 files, 6 dirs, 11004 bytes - targets 1/1
OK       folder       test_files_max /opt/cmt/testdata OK - 7 files, 6 dirs, 11004 bytes - targets 1/1
CRITICAL folder       test_size_min /opt/cmt/testdata : too small (11004)
CRITICAL folder       test_size_max /opt/cmt/testdata : too big (11004)
OK       folder       test_has_recent /opt/cmt/testdata OK - 7 files, 6 dirs, 11004 bytes - targets 1/1
OK       folder       test_has_old /opt/cmt/testdata OK - 7 files, 6 dirs, 11004 bytes - targets 1/1
OK       folder       test_missing /opt/cmt/testdata/file.txt OK - 1 files, 0 dirs, 0 bytes - targets 0/0
OK       folder       test_nostore /opt/cmt/testdata/file.txt OK - 1 files, 0 dirs, 0 bytes [0.0 B] - targets 0/0
SKIPPED  folder       folder_root must run as root
OK       folder       folder_list /opt/cmt OK - 7 files, 1 dirs, 27593 bytes - targets 0/0
OK       memory       mem used 50.2 % - used 1.7 GB - avail 2.1 GB - total 4.1 GB
OK       mount        mount / found
ERROR    mount        mount /mnt not found
OK       ping         ping ok
OK       ping         ping localhost ok
OK       ping         ping ok
CRITICAL ping         ping not responding
CRITICAL ping         ping not responding
CRITICAL process      process redis missing (redis, None)
CRITICAL process      process apache missing (httpd, None)
OK       process      process cron found (cron, None) - memory rss 3.0 MB - cpu 0.0 sec.
OK       process      process ssh found (sshd, None) - memory rss 5.3 MB - cpu 0.01 sec.
CRITICAL process      process ntp missing (ntpd, None)
OK       process      process mysql found (mysqld, None) - memory rss 96.0 MB - cpu 7.93 sec.
CRITICAL process      process php-fpm missing (php-fpm, None)
OK       socket       socket local redis localhost tcp/6379 - alive: yes - count: 0
OK       socket       socket remote www_google tcp/443 - alive: yes - count: 0
OK       swap         swap used: 0.2 % /  4.5 MB - total 2.1 GB
OK       url          url - [Host: ] - http=200 - 127 ms ; pattern OK
OK       url          url header_check - [Host: ] - http=200 - 64 ms ; pattern OK
CRITICAL url          url www_non_existing - http://www.nonexisting/ [Host: ] - timeout/no response to query
OK       url          url google - [Host: ] - http=200 - 102 ms ; pattern OK
OK       url          url yahoo - [Host: ] - http=200 - 751 ms ; pattern OK
SKIPPED  send         test_token1 must be run as single check w/ stdin piped command
OK       mysql        mydbmaster - cx=1 cx/s=0 r/s=0 w/s=0 q/s=0 mem=275195600
CRITICAL mysql        mysql - can't connect with conf /opt/cmt/mysql_slave.cnf
OK       sendfile     /opt/cmt/demo.json - 3 lines/events
OK       mysqldata    user - 3 lines collected

2022/01/23 - 16:54:44 : SEVERITY=CRITICAL - 44/60 OK (73 %) - 16 NOK : 14 criticial - 1 error - 1 warning - 0 notice.

Data model / data storage

CMT provides a clean and well organized data model to enable an efficient metrology storage and further processing of collected data. You can then build various dashborad for data analysis and correlation, for reporting, finer alerting, etc.

Deploy easily

CMT is provided as a standalone binary for most recent linux distributions. Download and get running. Optionnaly yon can download and run directly the python3 code with a few mandatory librairies (yaml, requests, psutils, ...)


Ultimately, due to its simple configuration and modular design, it is very easy to add new checks or new modules, thus enabling a single tool to cover technical and functionnal monitoring and alerting needs.

Command and shortcuts

  • cmt --help
  • (...)

CMT remote management

CMT optionally supports additional configuration items from a remote URL, downloaded at runtime. It enables various modification from a central server (enable or disable some modules and check, require additional checks, ...).

No assumption is made on the remote server providing additional configuration. It can be as simple as text files from a shared Web server. It can also be an API to your existing systems which have to respond for a given CMT node, with a slice of configuration to be merged in the node's local configuration.

Modules and Checks

Think of module has class of check for a specific family of items, like URL, folder, process, disk... Modules are internally implemented as small python file in the checks/ folder.

Checks are instance of module execution. Each check create one metrology message or event. A check can carry several datapoint (checkitems). Each checks carries optionnal alert,warning, notice information and a message formatted for human reading.

Each Check depending on his Module capacbilities can set information such as alert, warning, notice that will be sent back to metrology servers and possibly direct Pagers services.

Pager notifications are sent only once per CMT exectuion at most. It is intended for (very) standalone usage of CMT, whereas advanced usage with many nodes will benefit from Pager notification fired from central systms where decisions can be made from many/all monitoring sources.

CMT provides various configuration to ease configuring alert levels, frequency, rate-limit, max level, threshold, and other stop switches to garantee a clean thread of monitoring data to Metrology servers and Pagers.

CMT automated run from crontab

CMT is intended to run from crontab, every minute. It has a built-in max-duration security to prevent unfinished executing to overcrowd a server.

Deployment layout

/usr/local/bin/cmt  # binary standalone

/opt/cmt            # Home dir
    cmt             # cmt binary agent if prefered over /usr/local/bin
    mysql.cnf       # credentials for mysql checks if needed
    conf.yml        # configuration what to measure / where to send
    conf.d/         # additional configuration (eg. ansible additive deployment)
    persist.json    # optional data persistance across run (crontab mode mostly)

Why another monitoring tool ?

The initial requirements were :

  • small footprint agent, easy to deploy (drop anywhere ...)
  • usable both with CRON and as interactive/CLI directly on each system
  • clean metric data structure (multi-tenant, multi-customers, multi nodes)
  • simple configuration file (one for all)
  • data agnostic: technical, business, ...
  • push rather than pull : push data to metrology servers
  • data is sent to very standard ElasticSearch / INfluxDB DataStore, private or in the cloud
  • modular design for easy extension / additional checks or modules
  • minimal learning curve : drop the binary, edit the conf, run as cron/CLI, send to Elasticsearch, create Dashboard.
  • ansible aware for mass deployement and standardization
  • open source and hopefully well documented enough

None of the reviewed tools covered this exact set of requirements, some being too specific on technical metrics, others requiring a steep learning curve (eg. to extend), other used specific datastore and needed very specific protocol to be allowed (network/firewalls) between agents and datastore ...


CMT is released under BSD-3clause licence, free to use, free to extend. Experiment, Adopt, Enjoy, Contribute.