Mr. Poke

Pokes lazy people to review their Gitlab MRs.

ToC

• Mr. Poke
  • ToC
  • Description
  • Setup
    • Prerequisites
      • Postgres database
      • Slack App
      • GitLab App
      • GitLab group webhook
      • Mr. Poke service deployment
        • Bare metal
        • Docker
        • Helm chart
        • Terraform
  • TODOs

Description

The service allows to setup plain and simple slack notifications for GitLab. It doesn't annoy you with creating channels per MR, all notifications go to the slack app messages.

What does it do:

• Notifies MR authors when they forgot to put assignees and reviewers.
• Notifies reviewers when MR is opened (or updated from Draft state).
• Notifies asignees when MR was approved and became ready to merge.

In the same time it's 100% secure, since your data stays in your organisation and is not sent to any 3rd parties.

Setup

Prerequisites

Postgres database

Backend uses prisma to connect to database. For now it's set to postgres. But you may adjust it to your needs by changing datasource provider in prisma/schema.prisma.

Slack App

1. Create app in slack.
2. Go to "OAuth & Permissions" and add scopes. For now it's enough chat:write, users.profile:read, users:read, users:read.email. In future it may evolve with adding new functions.
3. Install your app into workspace.
4. Save your Bot User OAuth Token for later.

GitLab App

1. Go to your GitLab and create an app with scopes read_user, openid, profile, email.
2. Optionally add redirect URI, if you already know the domain where it will be hosted (example: https://mr-poke.example.com/login/gitlab/callback).
3. Save Application ID and Secret for later.

GitLab group webhook

1. Go to your GitLab and create an create groop webhook.
2. Put your url, for example https://mr-poke.example.com/hooks/gitlab and select "Merge request events".
3. Create "Secret token" for webhook and save it.

Mr. Poke service deployment

Bare metal

To run the service locally or on bare metal you need node.js version 18+ and pnpm.

1. Clone the repo
2. Install dependencies via pnpm i
3. Setup needed environment variables (see docker for the list)
4. Run via pnpm run start:prod

Docker

Docker images are available in repo: SkeLLLa/mr-poke

docker pull ghcr.io/skellla/mr-poke:latest

Available environment variables:

  DATABASE_URL="<postgres connection string>"
  GL_HOOK_TOKEN="<gitlab webhook secret token for request validation>"
  GL_CLIENT_ID="<gitlab application id>"
  GL_CLIENT_SECRET="<gitlab application secret>"
  OAUTH2_CALLBACK_URI="<gitlab oaut2 callback uri>"
  GL_ACCESS_TOKEN="<gitlab access token>"
  SLACK_TOKEN="<slack bot token>"

For other env variables see config.

Example:

  GL_CLIENT_ID="01....."
  GL_CLIENT_SECRET="gloas-...."
  OAUTH2_CALLBACK_URI="https://mr-poke.example.com/login/gitlab/callback"
  GL_ACCESS_TOKEN="glpat-...."
  SLACK_TOKEN="xoxb-...-...-..."
  GL_HOOK_TOKEN="super-secret"
  DATABASE_URL="postgresql://user:[email protected]:5432/mr-poke-db"
  SLACK_EMAIL_DOMAINS="example.com"
  PG_BOUNCER_ENABLED=false
  PG_SSL_MODE=disable
  LOG_LEVEL=debug
  SERVER_PORT=3000

Helm chart

Helm chart is available via OCI repo

To see available versions and default values use command:

 helm show all oci://ghcr.io/skellla/charts/mr-poke

To install app you can use helm command as well, but don't forget to create your values.yaml and pass env variables mentioned in docker section.

Example:

env:
  PG_BOUNCER_ENABLED: 'true'
  PG_SSL_MODE: 'require'
  ENV: 'production'
  NODE_ENV: 'production'
  LOG_LEVEL: 'debug'
  GL_CLIENT_ID: '01.....'
  GL_CLIENT_SECRET: 'gloas-....'
  OAUTH2_CALLBACK_URI: 'https://mr-poke.example.com/login/gitlab/callback'
  GL_ACCESS_TOKEN: 'glpat-....'
  SLACK_TOKEN: 'xoxb-...-...-...'
  GL_HOOK_TOKEN: 'super-secret'
  DATABASE_URL: 'postgresql://user:[email protected]:5432/mr-poke-db'
  SLACK_EMAIL_DOMAINS: 'example.com'

service:
  type: ClusterIP
  port: 80

ingress:
  enabled: true
  annotations:
    kubernetes.io/ingress.class: nginx
    nginx.org/mergeable-ingress-type: minion
    kubernetes.io/tls-acme: 'true'
    nginx.ingress.kubernetes.io/jaeger-service-name: 'ingress-service-mr-poke'
    nginx.ingress.kubernetes.io/configuration-snippet: |
      more_set_headers "server: testing";
      more_set_headers "X-Content-Type-Options: nosniff";
      more_set_headers "X-Frame-Options: DENY";
      more_set_headers "X-Xss-Protection: 1";
      more_set_headers "X-Robots-Tag: noindex, nofollow, nosnippet, noarchive";
  hosts:
    - host: ''
      paths:
        - /

affinity:
  podAntiAffinity:
    requiredDuringSchedulingIgnoredDuringExecution:
      - labelSelector:
          matchExpressions:
            - key: app.kubernetes.io/instance
              operator: In
              values:
                - service-mr-poke
        topologyKey: 'kubernetes.io/hostname'

serviceMonitor:
  enabled: true
  labels:
    prometheus: monitoring
    release: prometheus
  interval: 1m
  scrapeTimeout: 30s

 helm install mr-poke oci://ghcr.io/skellla/charts/mr-poke -f my-values.yaml

Terraform

Install using terraform and helm release provider. Example config:

variable "host" {
  type = string
}

variable "slack_notifications_db_uri" {
  sensitive = true
  type      = string
}

variable "gitlab_webhook_token" {
  sensitive = true
  type      = string
}

variable "gitlab_app_id" {
  sensitive = true
  type      = string
}

variable "gitlab_app_secret" {
  sensitive = true
  type      = string
}

variable "gitlab_oauth_callback_uri" {
  type = string
}

variable "slack_token" {
  type      = string
  sensitive = true
}

variable "gitlab_access_token" {
  type      = string
  sensitive = true
}

resource "helm_release" "mr_poke" {
  name             = "mr-poke"
  create_namespace = true
  cleanup_on_fail  = true
  namespace        = "default"
  repository       = "oci://ghcr.io/skellla/charts"
  chart            = "mr-poke"
  version          = "1.0.0"
  values = [
    "${file("${path.module}/mr-poke.values.yaml")}"
  ]

  set {
    name  = "ingress.hosts[0].host"
    value = var.host
  }

  set {
    name  = "ingress.tls[0].hosts[0]"
    value = var.host
  }

  set {
    name  = "env.DATABASE_URL"
    value = var.slack_notifications_db_uri
  }

  set {
    name  = "env.GL_HOOK_TOKEN"
    value = var.gitlab_webhook_token
  }

  set {
    name  = "env.GL_CLIENT_ID"
    value = var.gitlab_app_id
  }

  set {
    name  = "env.GL_CLIENT_SECRET"
    value = var.gitlab_app_secret
  }

  set {
    name  = "env.OAUTH2_CALLBACK_URI"
    value = var.gitlab_oauth_callback_uri
  }

  set {
    name  = "env.GL_ACCESS_TOKEN"
    value = var.gitlab_access_token
  }

  set {
    name  = "env.SLACK_TOKEN"
    value = var.slack_token
  }
}

TODOs

There are lot's of thing could be done. But hopefully GitLab team after years of development will make their slack app usable.

Here's a small list of possible improvements, so PRs welcome:

• [] Add periodic notification to reviewers if MR is not ready to merge (not enough approvals).
• [] Integrate i18next to localize slack messages based on user locale.
• [] Add application homepage with list of unreviewed MRs.
• [] Add npm binary package for those who don't want installing gitlab webhooks.
• [] You tell me.