Skip to content

Latest commit

 

History

History
158 lines (103 loc) · 7.49 KB

README.md

File metadata and controls

158 lines (103 loc) · 7.49 KB
Raw

calert

Send Alertmanager notifications to Google Chat (and more!)

calert uses Alertmanager webhook receiver to receive alerts payload, and pushes this data to Google Chat webhook endpoint.

Quickstart

Binary

Grab the latest release from Releases.

To run:

./calert.bin --config config.toml

Docker

You can find the list of docker images here

docker pull ghcr.io/mr-karan/calert:latest

Here's an example docker-compose config with a custom message.tmpl mounted inside the container:

  calert:
    image: ghcr.io/mr-karan/calert:latest
    ports:
      - "6000:6000"
    volumes:
      - ./message.tmpl:/etc/calert/message.tmpl

Configuration

Refer to config.sample.toml for instructions on how to configure calert.

All the config variables can also be supplied as Environment Variables by prefixing CALERT_ and replacing . (period) with __ (double underscores).

Example:

  • app.address would become CALERT_APP__ADDRESS

App

Key Explanation Default
app.address Address of the HTTP Server. 0.0.0.0:6000
app.server_timeout Server timeout for HTTP requests. 5s
app.enable_request_logs Enable HTTP request logging. true
app.log Use debug to enable verbose logging. Can be set to info otherwise. info

Providers

calert can load a map of different providers. The unique identifier for the provider is the room name. Each provider has it's own configuration, based on it's provider_type. Currently calert` supports Google Chat but can support arbitary providers as well.

Key Explanation Default
providers.<room_name>.type Provider type. Currently only google_chat is supported. google_chat
providers.<room_name>.endpoint Webhook URL to send alerts to. -
providers.<room_name>.max_idle_conns Maximum Keep Alive connections to keep in the pool. 50
providers.<room_name>.timeout Timeout for making HTTP requests to the webhook URL. 7s
providers.<room_name>.template Template for rendering a formatted Alert notification. static/message.tmpl
providers.<room_name>.thread_ttl Timeout to keep active alerts in memory. Once this TTL expires, a new thread will be created. 12h

Alertmanager Integration

  • Alertmanager has the ability of group similar alerts together and fire only one event, clubbing all the alerts data into one event. calert leverages this and sends all alerts in one message by looping over the alerts and passing data in the template. You can configure the rules for grouping the alerts in alertmanager.yml config. You can read more about it here.

  • Configure Alertmanager config file (alertmanager.yml) and give the address of calert web-server. You can refer to the official documentation for more details.

You can refer to the following config block to route webhook alerts to calert:

route:
    receiver: 'calert'
    group_wait: 30s
    group_interval: 60s
    repeat_interval: 15m
    group_by: ['room', 'alertName']

receivers:
    - name: 'calert'
      webhook_configs:
      - url: 'http://calert:6000/dispatch'

Threading Support in Google Chat

calert ships with a basic support for sending multiple related alerts under a same thread, working around the limitations by Alertmanager.

Alertmanager currently doesn't send any Unique Identifier for each Alert. The use-case of sending related alerts under the same thread is helpful to triage similar alerts and see all their different states (Firing, Resolved) for people consuming these alerts. calert tries to solve this by:

  • Use the fingerprint field present in the Alert. This field is computed by hashing the labels for an alert.
  • Create a map of active_alerts in memory. Add an alert by it's fingerprint and generate a random UUID.v4 and store that in the map (along with some more meta-data like startAt field).
  • Use ?threadKey=uuid query param while making a request to Google Chat. This ensures that all alerts with same fingerprint (=same labels) go under the same thread.
  • A background worker runs every hour which scans the map of active_alerts. It checks whether the alert's startAt field has crossed the TTL (as specified by thread_ttl). If the TTL is expired then the alert is removed from the map. This ensures that the map of active_alerts doesn't grow unbounded and after a certain TTL all alerts are sent to a new thread.

Prometheus Metrics

calert exposes various metrics in the Prometheus exposition format.

Here's a list of internal app metrics available at /metrics:

Name Description Data type
calert_uptime_seconds Uptime of app (in seconds). counter
calert_start_timestamp UNIX timestamp since the app was booted. gauge
calert_http_requests_total Number of HTTP requests, grouped with labels like handler. counter
calert_http_request_duration_seconds_{sum,count,bucket} Duration of HTTP request (in seconds). histogram
calert_alerts_dispatched_total Number of alerts dispatched to upstream providers, grouped with labels like provider and room. counter
calert_alerts_dispatched_duration_seconds_{sum,count,bucket} Duration to send an alert to upstream provider. histogram

It also exposes Go process metrics in addition to app metrics, which you can use to monitor the performance of calert.

Migrating from v1 to v2

A few notes on v2 migration:

Config schema changes

v2 is a complete rewrite from scratch and is a breaking release. The configuration has changed extensively. Please refer to latest config.sample.toml for a complete working example of the config.

Dry Run Mode

In case you're simply experimenting with calert config changes and you don't wish to send actual notifications, you can set dry_run=true in each provider.

Room Name for Google Chat

Apart from the config, calert now determines the room based on the receiver specified in Alertmanager config. Previously, the room was identified with ?room query parameter in each HTTP request. However, since the Alert payload contains the receiver name, it's better to extract this information from the labels instead.

Here's an example of how Alertmanager config looks like. Notice the value of receiver (prod_alerts) should match one of provider.<room_name> (eg provider.prod_alerts) in your config.toml):

receivers:
    - name: 'prod_alerts'
      webhook_configs:
      - url: 'http://calert:6000/dispatch'

Contribution

PRs on Feature Requests, Bug fixes are welcome. Feel free to open an issue and have a discussion first.

For deployment manifests like Helm, Kustomize, Nomad etc - they're placed under contrib folder and generally manintained by the community.

License

LICENSE