HomeDocumentationAPI Reference
Log In

Deploying to Kubernetes

This article describes setting up and maintaining PrivX on Kubernetes.

Setup on Kubernetes

  1. Create a Kubernetes cluster for PrivX. The cluster version must be 1.19 or later.

  2. Create the privx namespace:

    kubectl create namespace privx
  3. Set up storage for PrivX data. This involves creating a persistent volume (PV), and a persistent-volume claim (PVC).

    • For PrivX-configuration files: Create a PV and PVC with accessModes set to ReadWriteMany. The PVC should be named privx-claim, and located in the privx namespace.
  4. Set up PrivX-TLS and Authorizer secrets.

    privx-tls: TLS-certificate data, required later for the ingress load balancer. The type of the secret must be kubernetes.io/tls. The secret must be in the privx namespace. The required secret contents are:

    • tls.crt: The TLS certificate.
    • tls.key: The corresponding private key.
    • ca.crt (optional): The certificate of the CA that issued the TLS certificate.

    You may generate the privx-tls secret similarly to the following:

    kubectl create secret tls privx-tls --cert=<Path to cert file> --key=<Path to key file> -n privx

    privx-ca-secret: The Authorizer certificate, used to issue certificates for certificate-based authentication. The secret type must be Opaque. The secret must be in the privx namespace. The required secret contents are:

    • ca.crt: The CA certificate (can be valid for longer time).

    You may generate the privx-ca-secret secret similarly to the following:

    kubectl create secret generic privx-ca-secret --from-file=ca.crt=<Path to ca cert file> -n privx



    Ensure that the secrets and their contents are named exactly as described.

  5. Install the nginx-ingress controller, used for load balancing:
    For Kubernetes version 1.19:

     helm repo add bitnami-full-index https://raw.githubusercontent.com/bitnami/charts/archive-full-index/bitnami
     helm install \
         -n ingress --create-namespace \
         -f values-overrides/ingress.yaml \
         --version 7.6.6 \
         ingress bitnami-full-index/nginx-ingress-controller

    For Kubernetes > 1.19:

    helm repo add bitnami https://charts.bitnami.com/bitnami
    helm install \
        -n ingress --create-namespace \
        -f values-overrides/ingress.yaml \
        --version 9.3.0 \
        ingress bitnami/nginx-ingress-controller

    Up to date instructions on new releases and any breaking changes can be found in the original repo.

    You will then need to set up the nginx-ingress load balancer with an external IP, and to add a DNS record for pointing to that IP. For provider-specific setup instructions, see the nginx-ingress Installation Guide.



    The values-overrides/ingress.yaml overrides some of the default values used in the helm chart for the Ingress controller. The file contains extra configurations and headers that are essential for PrivX to work correctly.

  6. (Optional) We recommend installing Reloader to automatically update certificates upon renewals:

    helm repo add stakater https://stakater.github.io/stakater-charts
    helm repo update
    helm install secrets-watcher stakater/reloader



    In the reloader's absence, the trust has to be updated manually.

  7. For production environments we recommend setting up a dedicated PostgreSQL cluster with replication.

    For evaluation purposes only, you may create a database within the Kubernetes cluster as follows:

    1. Create the db namespace.

      kubectl create namespace db
    2. Create a PV and PVC with accessModes set to ReadWriteMany. The PVC should be named db-psql-claim, and located in the db namespace.

    3. Install the database in the db namespace.

      kubectl apply -f db-psql-deployment.yaml -n db
  8. Install PrivX

    Duplicate a file like values-overrides/privx.yaml for your environment and change the ingress.common.host to match the DNS Domain that was created earlier.

    By default, PrivX containers are run under uid and gid 5111. To run containers under a custom uid/gid, customize the following values in charts/privx/values.yaml:

    • podSecurityContext.fsGroup
    • podSecurityContext.runAsUser
    • securityContext.runAsUser

    Note that the uid/gid cannot be changed after initial setup.

    If a database external to the cluster is used, then update dbAddress to point to the address of the database. Default value is db-psql.db. Replace the placeholder value of the license in ms.licensemanager.licenseCode.prod.value.

    helm install \
        -f $VALUE_OVERRIDES \
        -n privx \
        $PRIVX_RELEASE_NAME charts/privx/



    Privx is inaccessible unless the host setting matches the DNS domain.

    After successful installation, PrivX should be accessible from the browser using the domain.

    The helm chart creates and installs the following:

    • Creates configmaps that store information about privx internal settings (Only done when installing
      the helm chart for the first time. Upgrades have no effect).

    • Runs the installer job, which creates all the necessary resources for PrivX
      to run (for example volume mounts for directories like /opt/privx/, creates db etc).
      (Only done when installing the helm chart for the first time.
      Upgrades have no effect).

    • Creates deployments and pods for all microservices.

    • Creates kubernetes services for each microservices.

    • Creates ingress resources for accessing pods and services from outside
      the cluster.

Upgrade on Kubernetes

  1. Determine the current Helm revision of your PrivX release:

    helm history --namespace privx privx

    The last entry in the command is the current revision, usually the one with the STATUS=deployed.

  2. Create a volume claim named privx-backup-claim in the privx namespace for PrivX backups.

    Stop and back up PrivX with the following (replace <new_privx_version> with the version number you are upgrading to):

    helm upgrade --history-max 0 \
        -f values-overrides/privx.yaml \
        -f charts/privx/migrations/<new_privx_version>/stage1.yaml \
        --wait privx -n privx charts/privx/

    This creates the backup to the volume claimed by privx-backup-claim. The backup directory is named according to the following format:


  3. At this point, back up the PrivX database according to your database vendor's instructions.

  4. After the database and PrivX are successfully backed up, upgrade PrivX with the following (replace <new_privx_version> with the major version number you are upgrading to):

    helm upgrade --history-max 0 \
        -f values-overrides/privx.yaml \
        -f charts/privx/migrations/<new_privx_version>/stage2.yaml \
        privx -n privx charts/privx/

    This completes upgrade by updating PrivX configurations and starting new versions of the PrivX microservices.

Rolling Back Failed Upgrade on Kubernetes

If upgrade fails at any point, follow these instructions to get back to the original revision of PrivX:

  1. Stop PrivX:

    helm upgrade --history-max 0 \
        -f values-overrides/privx.yaml \
        --set shutdown=true \
        --wait privx -n privx charts/privx/
  2. Restore the database from backup.

  3. In restore.yaml, set BACKUP_DIR to the name of your backup directory (named similarly to privx-backup-PPPPP_YYYY-MM-DD-hhmm_<old_privx_version>). Make sure the correct backup directory is used. The backup directory name can also be copied from the logs of the backup job by running the following command:

    kubectl logs -n privx <name-of-the-backup-pod>
  4. Restore PrivX settings using the following command (replace <new_privx_version> with the major version number of PrivX that you tried upgrading to):

    helm upgrade --history-max 0 \
        -f values-overrides/privx.yaml \
        -f charts/privx/migrations/<new_privx_version>/restore.yaml \
        privx -n privx charts/privx/
  5. Roll back to the original PrivX revision (replace <revision-number> with the revision obtained from the first upgrade step):

    helm rollback --namespace privx privx <revision-number>

    Your PrivX deployment should now be restored to the pre-upgrade version.