Skip to content

Service to discover active web servers for prometheus scraping

Notifications You must be signed in to change notification settings

fnbrco/prometheus-discovery

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

19 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

prometheus-discovery

Service to discover active web servers for Prometheus scraping

How does this work?

This is a service to listen for 'events' to signify when servers have been added or removed from the active cluster.
When this happens a request is sent to DigitalOcean's API to get a list of droplets with the tag defined in the config.

Once this has happened a targets.json (name can be customised in config) file is written in a format Prometheus understands and will automatically notice changes and begin scraping those nodes for data with the criteria defined for that prometheus instance, such as scrape path and interval.

Triggering a rewrite

This service is triggered via a web request sent to /trigger with either the header webhook-key or GET parameter key. The request must use a valid webhook key defined in config.webhookKeys.
For example:

curl http://localhost:5001/trigger?key=super-strong-auth-key-here

Or with a header:

curl -H "webhook-key: super-strong-auth-key-here" http://localhost:5001/trigger

This approach allows both internal & external monitoring tools to trigger a rewrite.

Config

This service can be configured to fetch more than one 'group' of servers and attach labels (used in Prometheus queries) relevant to them, for example: region or role.
It also allows for multiple 'webhook keys' to allow more than one external service to trigger a rewrite of the targets file.

See config/config.default.js for an example config file, without any API credentials.

Config: File Mode

The service can be configured to write targets within tags to separate files or all into one master file. This can be useful depending on the deployment setting, for example if you are using a proxy_url within Prometheus for some of your targets (as is the case with PushProx).
If using single mode the outputFileName value does not need to be set for each tag section, as the outputFileName of the first tag will be used. If using multi mode there must be a file name for each tag.

Config key: config.prometheus.fileMode
Available config options: single or multi

Config: Network & Key Type

You can decide on a per-tag basis the type of IPv4 address that is obtained from a droplet.

Config key: config.prometheus.searchTags.$.network
Available config options: public or private

Additionally you can decide whether the IP address or droplet name is used in the target string. This is useful in certain usecases such as when using PushProx which uses hostname resolution.

Config key: config.prometheus.searchTags.$.keyType
Available config options: ip or name

Config: No servers found

By default, if no servers are found for any of the search tags the rewrite action will fail and respond with an error, you can override this on a per-tag basis.
If you set this value to true in multi file mode, an empty file will be written. In single mode the tag will be skipped.

Config key: config.prometheus.searchTags.$.allowEmpty
Available config options: true or false