How to initialize mysql container when created on Kubernetes?

Question

I want to set initial data on MySQL of container. In docker-compose.yml, such code can create initial data when running container.

volumes:
  - db:/var/lib/mysql
  - "./docker/mysql/conf.d:/etc/mysql/conf.d"
  - "./docker/mysql/init.d:/docker-entrypoint-initdb.d"

However, how can I create initial data on Kubernetes when running?

Solution 1

According to the MySQL Docker image README, the part that is relevant to data initialization on container start-up is to ensure all your initialization files are mount to the container's /docker-entrypoint-initdb.d folder.

You can define your initial data in a ConfigMap, and mount the corresponding volume in your pod like this:

apiVersion: v1
kind: Pod
metadata:
  name: mysql
spec:
  containers:
  - name: mysql
    image: mysql        
    ports:
      - containerPort: 3306
    volumeMounts:
      - name: mysql-initdb
        mountPath: /docker-entrypoint-initdb.d
  volumes:
    - name: mysql-initdb
      configMap:
        name: mysql-initdb-config
---
apiVersion: v1
kind: ConfigMap
metadata:
  name: mysql-initdb-config
data:
  initdb.sql: |
    CREATE TABLE friends (id INT, name VARCHAR(256), age INT, gender VARCHAR(3));
    INSERT INTO friends VALUES (1, 'John Smith', 32, 'm');
    INSERT INTO friends VALUES (2, 'Lilian Worksmith', 29, 'f');
    INSERT INTO friends VALUES (3, 'Michael Rupert', 27, 'm');

Solution 2

First: create persistent volume that contains your SQL scripts

kind: PersistentVolume
apiVersion: v1
metadata:
  name: mysql-initdb-pv-volume
  labels:
    type: local
    app: mysql
spec:
  storageClassName: manual
  capacity:
    storage: 1Mi
  accessModes:
    - ReadOnlyMany
  hostPath:
    path: "/path/to/initdb/sql/scripts"
---
kind: PersistentVolumeClaim
apiVersion: v1
metadata:
  name: mysql-initdb-pv-claim
  labels:
    app: mysql
spec:
  storageClassName: manual
  accessModes:
    - ReadOnlyMany
  resources:
    requests:
      storage: 1Mi

Note: assume that you have your SQL scripts in /path/to/initdb/sql/scripts

Second: mount the volume to /docker-entrypoint-initdb.d

apiVersion: extensions/v1beta1
kind: Deployment
metadata:
  name: mysql
spec:
  replicas: 1
  template:
    metadata:
      labels:
        app: mysql
    spec:
      containers:
        - name: mysql
          image: mysql
          imagePullPolicy: "IfNotPresent"
          ports:
            - containerPort: 3306
          volumeMounts:
            - mountPath: /docker-entrypoint-initdb.d
              name: mysql-initdb
      volumes:
        - name: mysql-initdb
          persistentVolumeClaim:
            claimName: mysql-initdb-pv-claim

That's it.

Note: this applies to PostgreSQL too.

Comments

[Harry]

I have some sql file to create tables and insert data. And I want to use for initialization when creating container. This code seems nothing to do that.

[sfgroups]

As ivan responded you need to mount the sql files file system using PV, PVC

[Harry]

Though I set hostPath, it seems requiring absolute path. However I faced this error. error: code = 2 desc = failed to start container "ae00a8dc51d8895506881960a310c447b67643ccd088d164891a68f32f1‌​24634": Error response from daemon: mkdir /gcptmp: read-only file system

[ivan.sim]

@Harry I have updated my answer to use volume with ConfigMap. Using hostPath requires you to either run your container as a privileged container or modify the file permission on your host to be writable to the hostPath volume, which in most cases aren't necessary.

[Harry]

It seems to be possible that configured yaml file bacome messy if sql is big. In my current situation, I created extended mysql image including initial sql files. However, I learned many from you. Thanks!!

[Harry]

it seems nice!! I'll try it! Thanks. After that I'll feed back it!

[pizza]

When I add the configmap to the volumes field, I get spec.template.spec.volumes[0].configMap: Forbidden: may not specify more than 1 volume type * spec.template.spec.containers[0].volumeMounts[0].name: Not found: "mysql-initdb" Does anybody know what the problem is?? This is my config volumes: - name: mysql-persistent-storage configMap: name: mysql-initdb-config persistentVolumeClaim: claimName: mysql-volumeclaim

[reactor]

when I use this exact pod config and run skaffold on it, I noticed that the pod is still in ContainerCreating status after 13 minutes and was wondering how to get it going. Thank you!

[user658182]

The problem with this approach is that data mapped to the configMap is limited to 3M. Take this example: kubectl create configmap init-script --from-file=db.sql error: failed to create configmap: Request entity too large: limit is 3145728 The challenge , especially for those of us coming from a LAMP stack background, your db backup is always going to be larger than 3M for an existing application. I'm facing the same question - When deploying an existing app, how do I get my existing data, files and db data, into that deployed app? Preferably automatically as part of a CI / CD pipeline.

[user658182]

References: github.com/bitnami/charts/issues/8378 and stackoverflow.com/q/70297296/658182. If anyone has overcome this issue and can provide insight, please contribute to the linked issues.