/gatus

⛑ Gatus - Automated service health dashboard

Primary LanguageGoMIT LicenseMIT

Gatus

build Go Report Card codecov Go version Docker pulls Follow TwinProduction

A service health dashboard in Go that is meant to be used as a docker image with a custom configuration file.

I personally deploy it in my Kubernetes cluster and let it monitor the status of my core applications: https://status.twinnation.org/

Table of Contents

Features

The main features of Gatus are:

  • Highly flexible health check conditions: While checking the response status may be enough for some use cases, Gatus goes much further and allows you to add conditions on the response time, the response body and even the IP address.
  • Ability to use Gatus for user acceptance tests: Thanks to the point above, you can leverage this application to create automated user acceptance tests.
  • Very easy to configure: Not only is the configuration designed to be as readable as possible, it's also extremely easy to add a new service or a new endpoint to monitor.
  • Alerting: While having a pretty visual dashboard is useful to keep track of the state of your application(s), you probably don't want to stare at it all day. Thus, notifications via Slack, PagerDuty and Twilio are supported out of the box with the ability to configure a custom alerting provider for any needs you might have, whether it be a different provider or a custom application that manages automated rollbacks.
  • Metrics
  • Low resource consumption: As with most Go applications, the resource footprint that this application requires is negligibly small.
  • Service auto discovery in Kubernetes (ALPHA)

Usage

By default, the configuration file is expected to be at config/config.yaml.

You can specify a custom path by setting the GATUS_CONFIG_FILE environment variable.

Here's a simple example:

metrics: true         # Whether to expose metrics at /metrics
services:
  - name: twinnation  # Name of your service, can be anything
    url: "https://twinnation.org/health"
    interval: 30s     # Duration to wait between every status check (default: 60s)
    conditions:
      - "[STATUS] == 200"         # Status must be 200
      - "[BODY].status == UP"     # The json path "$.status" must be equal to UP
      - "[RESPONSE_TIME] < 300"   # Response time must be under 300ms
  - name: example
    url: "https://example.org/"
    interval: 30s
    conditions:
      - "[STATUS] == 200"

This example would look like this:

Simple example

Note that you can also add environment variables in the configuration file (i.e. $DOMAIN, ${DOMAIN})

Configuration

Parameter Description Default
debug Whether to enable debug logs false
metrics Whether to expose metrics at /metrics false
services List of services to monitor Required []
services[].name Name of the service. Can be anything. Required ""
services[].url URL to send the request to Required ""
services[].conditions Conditions used to determine the health of the service []
services[].insecure Whether to skip verifying the server's certificate chain and host name false
services[].interval Duration to wait between every status check 60s
services[].method Request method GET
services[].graphql Whether to wrap the body in a query param ({"query":"$body"}) false
services[].body Request body ""
services[].headers Request headers {}
services[].alerts[].type Type of alert. Valid types: slack, pagerduty, twilio, custom Required ""
services[].alerts[].enabled Whether to enable the alert false
services[].alerts[].failure-threshold Number of failures in a row needed before triggering the alert 3
services[].alerts[].success-threshold Number of successes in a row before an ongoing incident is marked as resolved 2
services[].alerts[].send-on-resolved Whether to send a notification once a triggered alert is marked as resolved false
services[].alerts[].description Description of the alert. Will be included in the alert sent ""
alerting Configuration for alerting {}
alerting.slack Configuration for alerts of type slack {}
alerting.slack.webhook-url Slack Webhook URL Required ""
alerting.pagerduty Configuration for alerts of type pagerduty {}
alerting.pagerduty.integration-key PagerDuty Events API v2 integration key. Required ""
alerting.twilio Settings for alerts of type twilio {}
alerting.twilio.sid Twilio account SID Required ""
alerting.twilio.token Twilio auth token Required ""
alerting.twilio.from Number to send Twilio alerts from Required ""
alerting.twilio.to Number to send twilio alerts to Required ""
alerting.custom Configuration for custom actions on failure or alerts {}
alerting.custom.url Custom alerting request url Required ""
alerting.custom.body Custom alerting request body. ""
alerting.custom.headers Custom alerting request headers {}
security Security configuration {}
security.basic Basic authentication security configuration {}
security.basic.username Username for Basic authentication Required ""
security.basic.password-sha512 Password's SHA512 hash for Basic authentication Required ""
disable-monitoring-lock Whether to disable the monitoring lock false

For Kubernetes configuration, see Kubernetes

Conditions

Here are some examples of conditions you can use:

Condition Description Passing values Failing values
[STATUS] == 200 Status must be equal to 200 200 201, 404, ...
[STATUS] < 300 Status must lower than 300 200, 201, 299 301, 302, ...
[STATUS] <= 299 Status must be less than or equal to 299 200, 201, 299 301, 302, ...
[STATUS] > 400 Status must be greater than 400 401, 402, 403, 404 400, 200, ...
[CONNECTED] == true Connection to host must've been successful true, false
[RESPONSE_TIME] < 500 Response time must be below 500ms 100ms, 200ms, 300ms 500ms, 501ms
[IP] == 127.0.0.1 Target IP must be 127.0.0.1 127.0.0.1 0.0.0.0
[BODY] == 1 The body must be equal to 1 1 {}, 2, ...
[BODY].user.name == john JSONPath value of $.user.name is equal to john {"user":{"name":"john"}}
[BODY].data[0].id == 1 JSONPath value of $.data[0].id is equal to 1 {"data":[{"id":1}]}
[BODY].age == [BODY].id JSONPath value of $.age is equal JSONPath $.id {"age":1,"id":1}
len([BODY].data) < 5 Array at JSONPath $.data has less than 5 elements {"data":[{"id":1}]}
len([BODY].name) == 8 String at JSONPath $.name has a length of 8 {"name":"john.doe"} {"name":"bob"}
[BODY].name == pat(john*) String at JSONPath $.name matches pattern john* {"name":"john.doe"} {"name":"bob"}

Placeholders

Placeholder Description Example of resolved value
[STATUS] Resolves into the HTTP status of the request 404
[RESPONSE_TIME] Resolves into the response time the request took, in ms 10
[IP] Resolves into the IP of the target host 192.168.0.232
[BODY] Resolves into the response body. Supports JSONPath. {"name":"john.doe"}
[CONNECTED] Resolves into whether a connection could be established true

Functions

Function Description Example
len Returns the length of the object/slice. Works only with the [BODY] placeholder. len([BODY].username) > 8
pat Specifies that the string passed as parameter should be evaluated as a pattern. Works only with == and !=. [IP] == pat(192.168.*)

NOTE: Use pat only when you need to. [STATUS] == pat(2*) is a lot more expensive than [STATUS] < 300.

Alerting

Configuring Slack alerts

alerting:
  slack: 
    webhook-url: "https://hooks.slack.com/services/**********/**********/**********"
services:
  - name: twinnation
    url: "https://twinnation.org/health"
    interval: 30s
    alerts:
      - type: slack
        enabled: true
        description: "healthcheck failed 3 times in a row"
        send-on-resolved: true
      - type: slack
        enabled: true
        failure-threshold: 5
        description: "healthcheck failed 5 times in a row"
        send-on-resolved: true
    conditions:
      - "[STATUS] == 200"
      - "[BODY].status == UP"
      - "[RESPONSE_TIME] < 300"

Here's an example of what the notifications look like:

Slack notifications

Configuring PagerDuty alerts

It is highly recommended to set services[].alerts[].send-on-resolved to true for alerts of type pagerduty, because unlike other alerts, the operation resulting from setting said parameter to true will not create another incident, but mark the incident as resolved on PagerDuty instead.

alerting:
  pagerduty: 
    integration-key: "********************************"
services:
  - name: twinnation
    url: "https://twinnation.org/health"
    interval: 30s
    alerts:
      - type: pagerduty
        enabled: true
        failure-threshold: 3
        success-threshold: 5
        send-on-resolved: true
        description: "healthcheck failed 3 times in a row"
    conditions:
      - "[STATUS] == 200"
      - "[BODY].status == UP"
      - "[RESPONSE_TIME] < 300"

Configuring Twilio alerts

alerting:
  twilio:
    sid: "..."
    token: "..."
    from: "+1-234-567-8901"
    to: "+1-234-567-8901"
services:
  - name: twinnation
    interval: 30s
    url: "https://twinnation.org/health"
    alerts:
      - type: twilio
        enabled: true
        failure-threshold: 5
        send-on-resolved: true
        description: "healthcheck failed 5 times in a row"
    conditions:
      - "[STATUS] == 200"
      - "[BODY].status == UP"
      - "[RESPONSE_TIME] < 300"

Configuring custom alerts

While they're called alerts, you can use this feature to call anything.

For instance, you could automate rollbacks by having an application that keeps tracks of new deployments, and by leveraging Gatus, you could have Gatus call that application endpoint when a service starts failing. Your application would then check if the service that started failing was recently deployed, and if it was, then automatically roll it back.

The values [ALERT_DESCRIPTION] and [SERVICE_NAME] are automatically substituted for the alert description and the service name respectively in the body (alerting.custom.body) as well as the url (alerting.custom.url).

If you have send-on-resolved set to true, you may want to use [ALERT_TRIGGERED_OR_RESOLVED] to differentiate the notifications. It will be replaced for either TRIGGERED or RESOLVED, based on the situation.

For all intents and purpose, we'll configure the custom alert with a Slack webhook, but you can call anything you want.

alerting:
  custom:
    url: "https://hooks.slack.com/services/**********/**********/**********"
    method: "POST"
    body: |
      {
        "text": "[ALERT_TRIGGERED_OR_RESOLVED]: [SERVICE_NAME] - [ALERT_DESCRIPTION]"
      }
services:
  - name: twinnation
    url: "https://twinnation.org/health"
    interval: 30s
    alerts:
      - type: custom
        enabled: true
        failure-threshold: 10
        success-threshold: 3
        send-on-resolved: true
        description: "healthcheck failed 10 times in a row"
    conditions:
      - "[STATUS] == 200"
      - "[BODY].status == UP"
      - "[RESPONSE_TIME] < 300"

Kubernetes (ALPHA)

WARNING: This feature is in ALPHA. This means that it is very likely to change in the near future, which means that while you can use this feature as you see fit, there may be breaking changes in future releases.

Parameter Description Default
kubernetes Kubernetes configuration {}
kubernetes.auto-discover Whether to enable auto discovery false
kubernetes.cluster-mode Cluster mode to use for authenticating. Supported values: in, out Required ""
kubernetes.service-template Service template. See services[] in Configuration Required nil
kubernetes.excluded-service-suffixes List of service suffixes to not monitor (e.g. canary) []
kubernetes.namespaces List of configurations for the namespaces from which services will be discovered []
kubernetes.namespaces[].name Namespace name Required ""
kubernetes.namespaces[].hostname-suffix Suffix to append to the service name before calling target-path Required ""
kubernetes.namespaces[].target-path Path that will be called on the discovered service for the health check ""
kubernetes.namespaces[].excluded-services List of services to not monitor in the given namespace []

Auto Discovery

Auto discovery works by reading all Service resources from the configured namespaces and appending the hostname-suffix as well as the configured target-path to the service name and making an HTTP call.

All auto-discovered services will have the service configuration populated from the service-template.

You can exclude certain services from the dashboard by using kubernetes.excluded-service-suffixes or kubernetes.namespaces[].excluded-services.

kubernetes:
  auto-discover: true
  # out: Gatus is deployed outside of the K8s cluster.
  # in: Gatus is deployed in the K8s cluster
  cluster-mode: "out"                                              
  excluded-service-suffixes:
    - canary
  service-template:
    interval: 30s
    conditions:
      - "[STATUS] == 200"
  namespaces:
    - name: default
      # If cluster-mode is out, you should use an externally accessible hostname suffix (e.g.. .example.com)
      # This will result in gatus generating services with URLs like <service-name>.example.com
      # If cluster-mode is in, you can use either an externally accessible hostname suffix (e.g.. .example.com)
      # or an internally accessible hostname suffix (e.g. .default.svc.cluster.local)
      hostname-suffix: ".default.svc.cluster.local"
      target-path: "/health"
      # If some services cannot be or do not need to be monitored, you can exclude them by explicitly defining them
      # in the following list.
      excluded-services:
        - gatus
        - kubernetes

Note that hostname-suffix could also be something like .yourdomain.com, in which case the endpoint that would be monitored would be potato.example.com/health, assuming you have a service named potato and a matching ingress to map potato.example.com to the potato service.

Deploying

See example/kubernetes-with-auto-discovery

Docker

Other than using one of the examples provided in the examples folder, you can also try it out locally by creating a configuration file - we'll call it config.yaml for this example - and running the following command:

docker run -p 8080:8080 --mount type=bind,source="$(pwd)"/config.yaml,target=/config/config.yaml --name gatus twinproduction/gatus

If you're on Windows, replace "$(pwd)" by the absolute path to your current directory, e.g.:

docker run -p 8080:8080 --mount type=bind,source=C:/Users/Chris/Desktop/config.yaml,target=/config/config.yaml --name gatus twinproduction/gatus

Running the tests

go test ./... -mod vendor

Using in Production

See the example folder.

FAQ

Sending a GraphQL request

By setting services[].graphql to true, the body will automatically be wrapped by the standard GraphQL query parameter.

For instance, the following configuration:

services:
  - name: filter users by gender
    url: http://localhost:8080/playground
    method: POST
    graphql: true
    body: |
      {
        user(gender: "female") {
          id
          name
          gender
          avatar
        }
      }
    headers:
      Content-Type: application/json
    conditions:
      - "[STATUS] == 200"
      - "[BODY].data.user[0].gender == female"

will send a POST request to http://localhost:8080/playground with the following body:

{"query":"      {\n        user(gender: \"female\") {\n          id\n          name\n          gender\n          avatar\n        }\n      }"}

Recommended interval

NOTE: This does not really apply if disable-monitoring-lock is set to true, as the monitoring lock is what tells Gatus to only evaluate one service at a time.

To ensure that Gatus provides reliable and accurate results (i.e. response time), Gatus only evaluates one service at a time In other words, even if you have multiple services with the exact same interval, they will not execute at the same time.

You can test this yourself by running Gatus with several services configured with a very short, unrealistic interval, such as 1ms. You'll notice that the response time does not fluctuate - that is because while services are evaluated on different goroutines, there's a global lock that prevents multiple services from running at the same time.

Unfortunately, there is a drawback. If you have a lot of services, including some that are very slow or prone to time out (the default time out is 10s for HTTP and 5s for TCP), then it means that for the entire duration of the request, no other services can be evaluated.

This does mean that Gatus will be unable to evaluate the health of other services. The interval does not include the duration of the request itself, which means that if a service has an interval of 30s and the request takes 2s to complete, the timestamp between two evaluations will be 32s, not 30s.

While this does not prevent Gatus' from performing health checks on all other services, it may cause Gatus to be unable to respect the configured interval, for instance:

  • Service A has an interval of 5s, and times out after 10s to complete
  • Service B has an interval of 5s, and takes 1ms to complete
  • Service B will be unable to run every 5s, because service A's health evaluation takes longer than its interval

To sum it up, while Gatus can really handle any interval you throw at it, you're better off having slow requests with higher interval.

As a rule of the thumb, I personally set interval for more complex health checks to 5m (5 minutes) and simple health checks used for alerting (PagerDuty/Twilio) to 30s.

Default timeouts

Protocol Timeout
HTTP 10s
TCP 5s

Monitoring a TCP service

By prefixing services[].url with tcp:\\, you can monitor TCP services at a very basic level:

  - name: redis
    url: "tcp://127.0.0.1:6379"
    interval: 30s
    conditions:
      - "[CONNECTED] == true"

Placeholders [STATUS] and [BODY] as well as the fields services[].body, services[].insecure, services[].headers, services[].method and services[].graphql are not supported for TCP services.

NOTE: [CONNECTED] == true does not guarantee that the service itself is healthy - it only guarantees that there's something at the given address listening to the given port, and that a connection to that address was successfully established.

Basic authentication

You can require Basic authentication by leveraging the security.basic configuration:

security:
  basic:
    username: "john.doe"
    password-sha512: "6b97ed68d14eb3f1aa959ce5d49c7dc612e1eb1dafd73b1e705847483fd6a6c809f2ceb4e8df6ff9984c6298ff0285cace6614bf8daa9f0070101b6c89899e22"

The example above will require that you authenticate with the username john.doe as well as the password hunter2.

disable-monitoring-lock

Setting disable-monitoring-lock to true means that multiple services could be monitored at the same time.

While this behavior wouldn't generally be harmful, conditions using the [RESPONSE_TIME] placeholder could be impacted by the evaluation of multiple services at the same time, therefore, the default value for this parameter is false.

There are three main reasons why you might want to disable the monitoring lock:

  • You're using Gatus for load testing (each services are periodically evaluated on a different goroutine, so technically, if you create 100 services with a 1 seconds interval, Gatus will send 100 requests per second)
  • You have a lot of services to monitor
  • You want to test multiple services at very short interval (< 5s)