loft-sh / component-chart Goto Github PK

The documentation currently only shows how to load single keys from Secrets and Configmaps using env.valueFrom.secretKeyRef and env.valueFrom.configMapKeyRef.

However, one can use envFrom.configMapRef and envFrom.secretRef to load all variables from a ConfigMap/Secret resp.:

yaml
      containers:
      - image: ${IMAGE}
        envFrom:
        - configMapRef:
            name: my-config-map

helm install -n my-namespace my-component https://charts.devspace.sh/component-chart

Error: failed to download "https://charts.devspace.sh/component-chart" (hint: running helm repo update may help)

version: v1beta11

vars:
- name: IMAGE
  value: username/app

deployments:
- name: test
  helm:
    componentChart: true
    values:
      initContainers:
      - image: ${IMAGE}
        volumeMounts:
          - containerPath: /foo
            volume:
              name: foo
      containers:
      - image: ${IMAGE}
      volumes:
        - name: foo
          secret:
            secretName: foo        

yields: (devspace render):

---
# Source: component-chart/templates/deployment.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
  name: "test"
  labels:
    "app.kubernetes.io/name": "devspace-app"
    "app.kubernetes.io/component": "test"
    "app.kubernetes.io/managed-by": "Helm"
  annotations:
    "helm.sh/chart": "component-chart-0.8.2"
spec:
  replicas: 1
  strategy:
    type: Recreate
  selector:
    matchLabels:
      "app.kubernetes.io/name": "devspace-app"
      "app.kubernetes.io/component": "test"
      "app.kubernetes.io/managed-by": "Helm"
  template:
    metadata:
      labels:
        "app.kubernetes.io/name": "devspace-app"
        "app.kubernetes.io/component": "test"
        "app.kubernetes.io/managed-by": "Helm"
      annotations:
        "helm.sh/chart": "component-chart-0.8.2"
    spec:
      imagePullSecrets:
      nodeSelector:
        null
      nodeName:
        null
      affinity:
        null
      tolerations:
        null
      dnsConfig:
        null
      hostAliases:
        null
      overhead:
        null
      readinessGates:
        null
      securityContext:
        null
      topologySpreadConstraints:
        null
      terminationGracePeriodSeconds: 5
      ephemeralContainers:
        null
      containers:
        - image: "username/app"
          name: "container-0"
          command:
          args:
          env:
            null
          envFrom:
            null
          securityContext:
            null
          lifecycle:
            null
          livenessProbe:
            null
          readinessProbe:
            null
          startupProbe:
            null
          volumeDevices:
            null
          volumeMounts:
      initContainers:
        - image: "username/app"
          name: "init-container-0"
          command:
          args:
          env:
            null
          envFrom:
            null
          securityContext:
            null
          lifecycle:
            null
          livenessProbe:
            null
          readinessProbe:
            null
          startupProbe:
            null
          volumeDevices:
            null
          volumeMounts:
            - mountPath: "/foo"
              name: "foo"
              readOnly: false
      volumes:
  volumeClaimTemplates:
---
# Source: component-chart/templates/deployment.yaml
# Create headless service for StatefulSet

version: componentChart 0.8.4
devspace cli version 5.17.0

In https://github.com/loft-sh/component-chart/blob/master/templates/pod-autoscaling.yaml it appears the HorizontalPodAutoscaler gets configured to target a StatefulSet if a deployment has any non-readonly volumes.

I am trying to create a deployment that supports s3fuse in EKS. Apparently I need podSecurityContext: privileged: true and serviceAccounts: [ ... ] for IRSA service account. It would seem that I can't use the component chart for this right now -- but it would be nice if I could! :)

\feature

I would like to have the possibility to use other metrics than those currently listed in the HPA resource file. I don't think it is very difficult to add some kind of user-defined metrics to your current HPA spec, is it?

I would like to use an emptyDir volume, which component-chart 0.8.4 supports. However, it seems that my deployments are using 0.8.2 (devspace version 5.17.0) and I can't figure out how to tell it to use a different version of the component chart.

When I add emptyDir I get the error:

[fatal]  deployments[6].helm.componentChart: component values are incorrect: yaml: unmarshal errors:
  line 49: field emptyDir not found in type latest.VolumeConfig

I tried removing emptyDir in order to deploy and then upgrade the chart, hoping I could then add back in. However, I get:

$ helm upgrade -n crane-web pgbouncer https://charts.devspace.sh/component-chart
Error: failed to download "https://charts.devspace.sh/component-chart"

In the service the app.kubernetes.io/name label is the name of the release. In the deployment it's devspace-app (and there's another label that points to the name of the release). It'd be less surprising if they were consistent, I think.

Hi, I'm deploying Ruby on Rails app with this chart, it requires running db migration to make sure db is up to date, usually I can define a Helm hook to start a job to update db.

This is an article describes how to deploy Rails app to K8s https://engineering.adwerx.com/rails-on-kubernetes-8cd4940eacbe , so I think most Rails apps need a job hook defined, can we add this feature to component-chart?

As of cert-manager version 1.6, supported annotations for cert-manager are as given here https://cert-manager.io/docs/usage/ingress/#supported-annotations
While component chart uses older version of annotations i.e

• certmanager.k8s.io/*

While expected for a StatefulSet, the volumeClaimsTemplate property is unexpected for a Deployment. Generating that property for a Deployment makes it hard to scan the rendered yaml manifests by devspace (e.g. with trivy or terrascan).

Hi All,

I have just started using DevSpace which seems really nice!!.

I currently migrating from Docker Compose with 10 services (Redis, MongoDB, NodeJS, Caddy etc).

I am trying to configure a MongoDB ReplicaSet Deployment using the Component Chart and our existing dockerfiles.

So far, I can get MongoDB to build and deploy fine on a K3D development cluster. The last remaining challenge is accessing the database server from outside the cluster.

From what I can tell, there appears to be no way to configure a nodePort for a service. You can set the type but cannot specify a port. It is always a random port.

If I try to add nodePort to the config, I get the following error - line 7: field nodePort not found in type latest.ServicePortConfig

So for now, I stopped using the component chart and switched to using manifests directly. Using manifests, I was able to set a nodePort on the service and can access the database from the host machine.

Is there a way to specify a nodePort?

Regards,
Tarek

I guess in Kubernetes 1.25 they removed “autoscaling/v2beta1”.
The error that I get is:
Error: UPGRADE FAILED: resource mapping not found for name: "my-release" no matches for kind "HorizontalPodAutoscaler" in version "autoscaling/v2beta1"

Can you please update the charts to support k8s v1.25+? Thank you

this might be entirely too advanced to reasonably request (and I can probably workaround with classic Helm charts)

I'm using vault-csi and secrets-store-csi to make Super Awesome k8s-Secrets integrations.

it requires a chunk volumes similar to this in order to instantiate Secrets Objects and perform the Vault API calls...

       volumes:
       - name: vault
         csi:
           driver: secrets-store.csi.k8s.io
           readOnly: true
           volumeAttributes:
             secretProviderClass: vault-api

predictably, it results in:

[fatal]  deployments[2].helm.componentChart: component values are incorrect: yaml: unmarshal errors:
  line 76: field csi not found in type latest.VolumeConfig
github.com/loft-sh/devspace/pkg/devspace/config/loader.validateDeployments

kubernetes-csi is some pretty new and powerful stuff, and I think it's The Way everything is going, sort of an "operating-system driver" metaphor for different types of vendor-specific or software-specific storage backends. (I believe most of the "built-in storage engines" are deprecated for CSI already.)

I did a quick look for csi in the repo and found nothing, so a quick smell-test on how possible and important it is before I keep attempting to hack on the problem. I'd rather contribute to Upstream and empower more people than workaround with classic Helm chart. perhaps the YAML smiles upon us today and extending VolumeConfig for csi is "fairly simple".

ImagePullPolicy is not documented in the docs containers section

As you can see here, helm is not going to add a namespace when we run helm template.
For this reason, many helm charts add a namespace: {{ .Release.Namespace }} to each manifest, so when running helm template/devspace render, the manifests include the namespace, so argocd or any other gitops like tool is able to work correctly.

Can you please add namespace: {{ .Release.Namespace }} ?

I am using component-chart with devspace. I want to deploy a PodMonitor into my stack but because I cannot define container ports there is no way to expose my metrics endpoint on Pod level. I am forced to use a ServiceMonitor instead.

Any chance this can be added ?

In the generated service, it sets ipFamily, but this seems like a typo and should be ipFamilyPolicy?

See https://kubernetes.io/docs/concepts/services-networking/dual-stack/

I'm deploying to a K3S cluster which has built-in Traefik ingress controller. It would be great to see an example of how that is configured in the component chart.

Hello,

When using wildcard domains in the host section, during the deployment there is a parsing issue:

rules:
  - host: "*.domain.tld"

[fatal]  Error deploying api: Unable to deploy helm chart: YAML parse error on component-chart/templates/ingress.yaml: error converting YAML to JSON: yaml: line 33: did not find expected alphabetic or numeric character

This should be fixed by adding quotes around the domain in the helm chart. See PR #31

For some usecases it would be great to add ConfigMaps from inline in devspace.yaml, so they can immediately used as volume in the containers.

It is currently possible to do this using kustomize's configMapGenerator along with helmChartInflationGenerator inflating the component-chart. However, this requires an additional file and it would be great to have it all completely inline with devspace.yaml.

One great way to implement this, would be to directly inject files as explained here: https://helm.sh/docs/chart_template_guide/accessing_files/#configmap-and-secrets-utility-functions

For Horizontal Pod Autoscaling,

While using autoscaling based on targetAverageUtilization, in was facing this issue in DevSpace version : 6.0.1

Input:

  replicas: 1
  autoScaling:
    horizontal:  
      maxReplicas: 3
      averageRelativeMemory: 40

Error:

ValidationError(HorizontalPodAutoscaler.spec.metrics[0].resource.targetAverageUtilization): invalid type for io.k8s.api.autoscaling.v2beta1.ResourceMetricSource.targetAverageUtilization: got "string", expected "integer"