- dj-wasabi.telegraf
This role will install and configure telegraf.
Telegraf is an agent written in Go for collecting metrics from the system it's running on, or from other services, and writing them into InfluxDB.
Design goals are to have a minimal memory footprint with a plugin system so that developers in the community can easily add support for collecting metrics from well known services (like Hadoop, Postgres, or Redis) and third party APIs (like Mailchimp, AWS CloudWatch, or Google Analytics).
(https://github.com/influxdb/telegraf)
This role supports the following systems:
- Red Hat
- Debian
- Ubuntu
- Docker container
- (Open)Suse
- Windows (Best effort)
- FreeBSD (Best effort)
So, you'll need one of those systems.. :-) Please sent Pull Requests or suggestions when you want to use this role for other systems.
You'll need an InfluxDB instance running somewhere on your network. Or 1 of the other output types found on https://github.com/influxdata/telegraf/#output-plugins
Docker needs to be installed on the target host. I can recommend these roles to install Docker:
This is only the case when the configuration is needed for a Telegraf inside a Docker container (When telegraf_agent_docker: True
).
There was an issue:
If I configure a telegraf_plugins_extra, run ansible, delete the plugin and run ansible again, the plugin stays on the machine.
Specifying the version to be installed:
telegraf_agent_version
: The version of Telegraf to install. Iftelegraf_agent_package_state
is set tolatest
, then this property and value is ignored. Default:1.10.0
How Telegraf
needs to be installed. There are 4 methods in getting Telegraf
installed on the target host:
- Via the package manager, like
yum
,apt
orzypper
("repo"); - Via a download from the
https://dl.influxdata.com/
site ("online"); - Already provided and is already available on the target host, but not yet installed/configured ("offline");
- Already installed on the target host or done manually, but not yet configured ("manual");
This can be configured by setting telegraf_agent_package_method
to one of the appropriate values ( repo
, online
, offline
or manual
).
These properties set in how and what package will be installed.
telegraf_agent_package
: The name of the Telegraf package to install. Whentelegraf_agent_package_method
is set toonline
oroffline
, it needs to have the full path of the file. Example:telegraf_agent_package: /tmp/telegraf.rpm
. Default:telegraf_agent_package: telegraf
.telegraf_agent_package_method
: The installation method to be used. Can choose between:repo
,offline
oronline
.telegraf_agent_package_state
: If the package should bepresent
orlatest
. When set tolatest
,telegraf_agent_version
will be ignored. Default:present
telegraf_agent_interval
: The interval configured for sending data to the server. Default:10
telegraf_agent_debug
: Run Telegraf in debug mode. Default:False
telegraf_agent_round_interval
: Rounds collection interval to 'interval' Default: Truetelegraf_agent_flush_interval
: Default data flushing interval for all outputs. Default: 10telegraf_agent_flush_jitter
: Jitter the flush interval by a random amount. Default: 0telegraf_agent_aws_tags
: Configure AWS ec2 tags into Telegraf tags section Default:False
telegraf_agent_aws_tags_prefix
: Define a prefix for AWS ec2 tags. Default:""
telegraf_agent_collection_jitter
: Jitter the collection by a random amount. Default: 0 (since v0.13)telegraf_agent_metric_batch_size
: The agent metric batch size. Default: 1000 (since v0.13)telegraf_agent_metric_buffer_limit
: The agent metric buffer limit. Default: 10000 (since v0.13)telegraf_agent_quiet
: Run Telegraf in quiet mode (error messages only). Default:False
(since v0.13)telegraf_agent_logfile
: The agent logfile name. Default: '' (means to log to stdout) (since v1.1)telegraf_agent_omit_hostname
: Do no set the "host" tag in the agent. Default:False
(since v1.1)
telegraf_agent_docker
: Install Telegraf as a docker container. Default:False
telegraf_agent_docker_name
: Name of the docker container. Default:telegraf
telegraf_agent_docker_network_mode
: Networking mode of the docker container. Default:bridge
telegraf_agent_docker_restart_policy
: Docker container restart policy. Default:unless-stopped
telegraf_uid_docker
: Override user id. Default:995
telegraf_gid_docker
: Override group id. Default:998
Full agent settings reference: https://github.com/influxdata/telegraf/blob/master/docs/CONFIGURATION.md#agent-configuration.
Extra info regarding: ansible_fqdn problematic for getting hostname #105
Describe the bug
In some nodes I'm getting weird hostnames, mostly localhost.localdomain. Those nodes show proper configuration in hostnamectl. I've seen you're using 'ansible_fqdn' as default.
Seems like ansible_fqdn and ansible_hostname can give different results, and sometimes even very weird results, as it sometimes makes DNS calls (which is not under my control in that cases) to infer that names.
Fix proposal
In my playbook I've added this parameter:
telegraf_agent_hostname: "{{ ansible_nodename }}"
You can set tags for the host running telegraf:
telegraf_global_tags:
- tag_name: some_name
tag_value: some_value
Specifying an output. The default is set to localhost, you'll have to specify the correct influxdb server:
telegraf_agent_output:
- type: influxdb
config:
- urls = ["http://localhost:8086"]
- database = "telegraf"
tagpass:
- cpu = ["cpu0"]
The config will be printed line by line into the configuration, so you could also use:
config:
- # Print an documentation line
and it will be printed in the configuration file.
The official Influxdata Telegraf image is used. telegraf_agent_version
will translate to the image tag.
Please note that the Docker container bind mounts basicly your whole system (read-only) to monitor the Docker Engine Host from within the container. To be precise:
- /etc/telegraf:/etc/telegraf:ro
- /:/hostfs:ro
- /etc:/hostfs/etc:ro
- /proc:/hostfs/proc:ro
- /sys:/hostfs/sys:ro
- /var/run:/var/run:ro
More information: https://github.com/influxdata/telegraf/blob/master/docs/FAQ.md.
telegraf_agent_docker: True
# Force host networking mode, so Docker Engine Host traffic metrics can be gathered.
telegraf_agent_docker_network_mode: host
# Force a specific image tag.
telegraf_agent_version: 1.10.0-alpine
telegraf_plugins_default:
- plugin: cpu
config:
- percpu = true
- plugin: disk
tagpass:
- fstype = [ "ext4", "xfs" ]
tagdrop:
- path = [ "/etc", "/etc/telegraf", "/etc/hostname", "/etc/hosts", "/etc/resolv.conf" ]
- plugin: io
- plugin: mem
- plugin: system
- plugin: swap
- plugin: netstat
- plugin: processes
- plugin: docker
config:
- endpoint = "unix:///var/run/docker.sock"
- timeout = "5s"
NOTE
Supporting Windows is an best effort (I don't have the possibility to either test/verify changes on the various amount of available Windows instances). PR's specific to Windows will almost immediately be merged, unless some one is able to provide a Windows test mechanism via Travis or other service for Pull Requests.
telegraf_win_install_dir
: The directory where Telegraf will be installed.telegraf_win_logfile
: The location to the logfile of Telegraf.telegraf_win_include
: The directory that will contain all plugin configuration.
NOTE
MacOS support is as the Window Support an best effort and not officially supported.
telegraf_mac_user
: Telegraf will run as this user (needed as running things as other users using brew is problematic)
There are two properties which are similar, but are used differently. Those are:
telegraf_plugins_default
telegraf_plugins_extra
With the property telegraf_plugins_default
it is set to use the default set of Telegraf plugins. You could override it with more plugins, which should be enabled at default.
telegraf_plugins_default:
- plugin: cpu
config:
- percpu = true
- plugin: disk
- plugin: io
- plugin: mem
- plugin: system
- plugin: swap
- plugin: netstat
Every telegraf agent has these as a default configuration.
The 2nd parameter telegraf_plugins_extra
can be used to add plugins specific to the servers goal. It is a hash instead of a list, so that you can merge values from multiple var files together. Following is an example for using this parameter for MySQL database servers:
cat group_vars/mysql_database
telegraf_plugins_extra:
mysql:
config:
- servers = ["root:{{ mysql_root_password }}@tcp(localhost:3306)/"]
There is an option to delete extra-plugin files in /etc/telegraf/telegraf.d if they weren't generated by this playbook with telegraf_plugins_extra_exclusive
variable.
Telegraf plugin options:
tags
An k/v tags to apply to a specific input's measurements. Can be used on any stage for better filtering for example in outputs.tagpass
: (added in Telegraf 0.1.5) tag names and arrays of strings that are used to filter metrics by the current plugin. Each string in the array is tested as an exact match against the tag name, and if it matches the metric is emitted.tagdrop
: (added in Telegraf 0.1.5) The inverse of tagpass. If a tag matches, the metric is not emitted. This is tested on metrics that have passed the tagpass test.interval
: How often to gather this metric. Normal plugins use a single global interval, but if one particular plugin should be run less or more often, you can configure that here.filter.name
: Like when there is an extra filter that needs to be configured, likegrok
for alogparser
plugin.filter.config
: The extra configuration for the - in thefilter.name
example -grok
filter. (See example below)
An example might look like this:
telegraf_plugins_default:
- plugin: disk
interval: 12
tags:
- diskmetrics = "true"
tagpass:
- fstype = [ "ext4", "xfs" ]
- path = [ "/opt", "/home" ]
If you want to define processors you can simply use telegraf_processors
variable.
An example might look like this:
telegraf_processors:
- processor: rename
- processor: rename.replace
config:
- tag = "level"
- dest = "LogLevel"
When you want to make use of the grok
filter for the logparser:
telegraf_plugins_extra:
logparser:
plugin: logparser
config:
- files = ["/var/log/messages"]
- from_beginning = false
filter:
name: grok
config:
- patterns = ["invoked oom-killer"]
No dependencies
- hosts: servers
roles:
- { role: dj-wasabi.telegraf }
This roles is configured to be tested with Molecule. You can find on this page some more information regarding Molecule: https://werner-dijkerman.nl/2016/07/10/testing-ansible-roles-with-molecule-testinfra-and-docker/
BSD
Please let me know if you have issues. Pull requests are also accepted! :-)
mail: ikben [ at ] werner-dijkerman . nl