Basic Datasette in Kubernetes#

This recipe for deploying the official datasetteproject/datasette container in Kubernetes just worked for me. It uses an interesting (possibly nasty?) trick to install plugins and download a SQLite database file on container startup, without needing to bake a brand new container image.

I’m running 2 replicas, to experiment with the cluster.

1
apiVersion: apps/v1
2
kind: Deployment
3
metadata:
4
  name: datasette-deployment
5
spec:
6
  replicas: 2
7
  selector:
8
    matchLabels:
9
      app: datasette
10
  template:
11
    metadata:
12
      labels:
13
        app: datasette
14
    spec:
15
      containers:
16
      - name: datasette
17
        image: datasetteproject/datasette
18
        command:
19
        - sh
20
        - -c
21
        args:
22
        - |-
23
          # Install some plugins
24
          pip install \
25
            datasette-debug-asgi \
26
            datasette-cluster-map \
27
            datasette-psutil
28
          # Download a DB (using Python because curl/wget are not available)
29
          python -c 'import urllib.request; urllib.request.urlretrieve("https://global-power-plants.datasettes.com/global-power-plants.db", "/home/global-power-plants.db")'
30
          # Start Datasette, on 0.0.0.0 to allow external traffic
31
          datasette -h 0.0.0.0 /home/global-power-plants.db
32
        ports:
33
        - containerPort: 8001
34
          protocol: TCP

The devious trick here is to use command and args to specify a multi-line shell script to run on startup, which installs some plugins with pip install, then downloads a database file using urllib.request.urlretrieve() in Python (because wget and curl are not included in the official image), then starts Datasette against that downloaded file.