Install on Kubernetes (Helm)

For production: two or more replicas, rolling upgrades, and a Postgres you already run.

What you need

Kubernetes 1.25 or newer, and Helm 3.
A Postgres database (14 or newer) and a user that owns it. The chart does not bring one: use CloudNativePG, your cloud's managed Postgres, or the server your other services use. Shorthop creates and migrates its own tables.

1. A Secret

kubectl create namespace shorthop
kubectl -n shorthop create secret generic shorthop \
  --from-literal=SECRET_KEY="$(openssl rand -hex 32)" \
  --from-literal=POSTGRES_PASSWORD='the database password'

Optional keys in the same Secret: RELAY_KEY, SMTP_PASSWORD, GOOGLE_CLIENT_SECRET, SLACK_BOT_TOKEN, SLACK_APP_TOKEN, METRICS_TOKEN.

2. Values

# values.yaml
postgres:
  host: postgres-rw.databases.svc
  database: shorthop
  user: shorthop
  sslmode: require

secret:
  existingSecret: shorthop

env:
  BOOTSTRAP_ADMIN_EMAIL: you@example.com
  PUBLIC_URL: https://oncall.internal.example.com
  PUBLIC_API_URL: https://oncall.example.com     # the public listener, for phones

ingress:                  # the web UI — an INTERNAL ingress class
  enabled: true
  className: internal
  host: oncall.internal.example.com

publicIngress:            # only what phones need — the one thing on the internet
  enabled: true
  className: public
  host: oncall.example.com

3. Install

helm install shorthop oci://ghcr.io/shorthop-io/charts/shorthop -n shorthop -f values.yaml
kubectl -n shorthop exec -it deploy/shorthop -- python -m app.cli admin you@example.com

The second command sets the first admin's password.

Replicas

replicaCount: 2 is the default and is safe: every pod sends notifications from a shared queue (nothing is sent twice), exactly one pod runs housekeeping at a time and another takes over within a tick, and the database migration on start takes a lock. A PodDisruptionBudget keeps one pod up during node drains. More replicas spread the load; they never change behaviour.

Metrics

serviceMonitor.enabled: true creates a ServiceMonitor for the Prometheus Operator. To alert on Shorthop itself, see Running it → monitoring Shorthop.

Upgrading

helm upgrade shorthop oci://ghcr.io/shorthop-io/charts/shorthop -n shorthop -f values.yaml

Pods roll one at a time; the first new pod migrates the database. Back up first.

All values: values.yaml.

Edit this page on GitHub