"Please provide a valid cache path" when running on K8s

[victorlane]

Hi all,

I'm trying to run Solidtime on my K8s cluster. I've run the app on docker-compose fine for a while. I've created the kubernetes resources to match my docker-compose setup however when running a container I constantly get the following error;

Defaulted container "app" out of: app, scheduler, queue, gotenberg, fix-permissions (init)
Container mode: http
Octane Server: frankenphp
Auto database migration enabled.

In Compiler.php line 67:
                                      
  Please provide a valid cache path.  

From what I gather this is a Laravel cache issue. I don't get what's causing it as I've not set any settings related to caching on any installation method.

These are my manifests (abbreviated to the app container only for now, the same error occurs across app, queue and scheduler).

# Storage
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: app-storage
  namespace: solidtime
spec:
  accessModes:
  - ReadWriteOnce
  storageClassName: local-path
  resources:
    requests:
      storage: 10Gi
---
apiVersion: apps/v1
kind: Deployment
metadata:
  name: solidtime
  namespace: solidtime
spec:
  replicas: 1
  selector:
    matchLabels:
      app: solidtime
  template:
    metadata:
      labels:
        app: solidtime
    spec:
      initContainers:
      - name: fix-permissions
        image: busybox:1.35
        command: [ 'sh', '-c' ]
        args:
        - |
          mkdir -p /var/www/html/storage/logs /var/www/html/storage/app
          chown -R 1000:1000 /var/www/html/storage
          chmod -R 755 /var/www/html/storage
        volumeMounts:
        - name: app-storage
          mountPath: /var/www/html/storage
        securityContext:
          runAsUser: 0
      containers:
      - name: app
        image: solidtime/solidtime:0.6
        ports:
        - containerPort: 8000
        envFrom:
        - secretRef:
            name: solidtime-env
        env:
        - name: CONTAINER_MODE
          value: "http"
        - name: AUTO_DB_MIGRATE
          value: "true"
        securityContext:
          runAsUser: 1000
          runAsGroup: 1000
        volumeMounts:
        - name: app-storage
          mountPath: /var/www/html/storage
        - name: app-storage
          mountPath: /var/www/html/storage/logs
          subPath: logs
        - name: app-storage
          mountPath: /var/www/html/storage/app
          subPath: app
        readinessProbe:
          httpGet:
            path: /health-check/up
            port: 8000
          initialDelaySeconds: 5
          periodSeconds: 10

      volumes:
      - name: app-storage
        persistentVolumeClaim:
          claimName: app-storage

With a pretty basic env setup:

APP_NAME="solidtime"
VITE_APP_NAME="solidtime"
APP_ENV="production"
APP_DEBUG="false"
APP_URL="https://<mydomain>"
APP_FORCE_HTTPS="true"
TRUSTED_PROXIES="0.0.0.0/0,2000:0:0:0:0:0:0:0/3"

# Authentication
APP_KEY="base64:<appkey>"
PASSPORT_PRIVATE_KEY=""
PASSPORT_PUBLIC_KEY=""
SUPER_ADMINS="<myuser>"

# Logging
LOG_CHANNEL="daily"
LOG_LEVEL="debug"

# Database connection settings abbreviated
# Mail abbreviated

# Queue
QUEUE_CONNECTION="database"

# File storage
FILESYSTEM_DISK="local"
PUBLIC_FILESYSTEM_DISK="public"

# Services
GOTENBERG_URL="http://gotenberg:3000"

Not sure what setting I am missing to get it to work.

[korridor]

If you use kubernetes I would advice you to use the database or redis cache driver. Since you already have a managed database and you are likely not scaling like crazy the database cache driver is likely the best choice.
See more infos here: https://laravel.com/docs/12.x/cache#main-content

[victorlane]

If you use kubernetes I would advice you to not use the database or redis cache driver. Since you already have a managed database and you are likely not scaling like crazy the database cache driver is likely the best choice. See more infos here: https://laravel.com/docs/12.x/cache#main-content

Sorry for the delayed reply, gonna try this today. Will close the issue afterwards! :)

[victorlane]

Setting the CACHE_STORE and CACHE_DRIVER env variables fixed my issue, alongside passing the DB connection variables to the queue and scheduler containers.

[korridor]

Glad to hear. I just saw that my typed something wrong in my last comment. (I edited it now) Please DO USE the database or redis cache driver and NOT the file driver. The file driver on kubernetes is not a good idea since it will do very weird stuff if you have more than one pod and no network storage.