esttorhe / stethoscope

External websites monitoring tool

stethoscope

External websites monitoring tool

Description

This is a very naive service that will load a set of Rules from a configuration file (rules.yml) that will later use to monitor the «heartbeat» of the websites.

Composition

Counter

A counter is the definition of a Prometheus counter that will be used when a monitored website is down. They are loaded from counters.yml and are defined as so:

   - namespace: website
     subsystem: health
     name: page_down
     help: Tracks the number of times there's an error loading a website
     labels: 
       - 'website'
       - 'status_code'

Field	Description
namespace	Prometheus namespace is an application prefix relevant to the domain the metric belongs to.1
subsystem	Prometheus subsystem is used to prepend the counter; it is prepended after the namespace
name	The identification name for the counter. This will also be the mapping key that should match the counter specified in the rules.
help	Description of what this counter is and what is it tracking.
labels	Array of labels that will be added to the counter. Right now it always expects this 2 website & status_code but will make them configurable in the future.

Rule

A rule is a basic definition o what should be monitored and how. It can be defined like this:

  - name: name
    website: website
    counter: counter_name
    interval: 3600000000000 # Check every hour
    timeout: 2000000000 # Timeout after 2 seconds
    use_head: true

Field	Description
name	The name of your rule. Used mainly for logging purposes
website	An environment variable that contains the value of the website you want to monitor. This was done like this in order to OSS the project without exposing the websites I'm monitoring.
counter	The name of the counter that should be used when the website is down. Should be a match with the name field in the one of the counters from counters.yml
interval	Number of nanoseconds to wait before attempting to monitor the website.
timeout	Number of nanoseconds to wait for the website before considering it as down (due to a timetout)
use_head	If the monitoring should be attempted using a HEAD method. If false the service will use GET.

Docker

To make my life easier I created a [Dockerfile][docker] that compiles the service into a binary using a Go 1.12 image. Once the binary is created it uses a custom Alpine image with CA certificates to wrap the binary and uses it as the base for the executing image.

All in all pretty simple but I find it quite comfortable to build on a container (which gives me confidence that it will work since technically it will always use the same environment for building).

Running it

Once the image is ready that «recommended» way of executing ti would be like so:

docker run -dt --name=stethoscope -p 7000:7000 stethoscope:latest

This obviously if you are running it locally; otherwise you need to specify your registry.

Makefile

Because I'm also lazy I created a Makefile whose sole purpose is to compile the service, generate the Docker image and upload it to the registry. This way I can ssh into my server and just pull the image and launch a new container with the latest image.

Easy peasy!

Author

Esteban Torres

• 
• ✉ [e-mail]

License

stethoscope is available under the MIT license. See the LICENSE file for more info.

Footnotes

1. https://prometheus.io/docs/practices/naming/#metric-names [docker]:./Dockerfile ↩