Operator based Installation

K10 Operator Editions

  • Kasten K10 (Free): Free edition of K10 for use in clusters up to 5 nodes

  • Kasten K10 (Enterprise - PAYGO): Enterprise edition of K10, billed per usage of node-hours

  • Kasten K10 (Enterprise - Term): Enterprise edition of K10 intended to be used with a term license

Pre-Flight Checks

Assuming that your default oc context is pointed to the cluster you want to install K10 on, you can run pre-flight checks by deploying the primer tool. This tool runs in a pod in the cluster and does the following:

  • Validates if the Kubernetes settings meet the K10 requirements.

  • Catalogs the available StorageClasses.

  • If a CSI provisioner exists, it will also perform a basic validation of the cluster's CSI capabilities and any relevant objects that may be required. It is strongly recommended that the same tool be used to also perform a more complete CSI validation using the documentation here.

Note that this will create and clean up a ServiceAccount and ClusterRoleBinding to perform sanity checks on your Kubernetes cluster.

Run the following command to deploy the pre-check tool:

$ curl https://docs.kasten.io/tools/k10_primer.sh | bash

Prerequisites

Before installing Kasten K10, it is essential to have a functional and accessible Red Hat OpenShift environment.

Optionally, you can create a new project in advance where Kasten K10 will be installed. Select this project during operator deployment or create a project (namespace) during the operator installation process. By default, the documentation uses the kasten-io namespace.

oc new-project kasten-io \
  --description="Kubernetes data management platform" \
  --display-name="Kasten K10"

K10 Install

Interactive Demo


Step-by-Step Guide

  1. Select the OperatorHub from the Operators Menu, search for Kasten. Select either the Certified Operator, or Marketplace version depending on the requirements

  1. To begin the installation, simply click Install

  1. Next, set the channel to stable and installation mode to A specific namespace on the cluster. Choose the kasten-io project created in an earlier step.

  1. Once installed, a K10 instance can be created by clicking Create Instance in the operator details page.

  1. The default installation can be done through either the Form View or YAML View. There are no changes needed to install by default.

    K10 assumes that the default storage class is supported by SSDs or equivalent fast storage media. If this assumption is not true, please modify the installation values to specify a performance-oriented storage class. This modification can be done within the form view or directly within the YAML of the K10 Operand configuration by setting the parameters below:

    global:
  persistence:
    storageClass: <storage-class-name>

Offline Operator Install

Note

Only K10 Operator Editions "Kasten K10 (Free)" and "Kasten K10 (Enterprise - Term)" are supported in disconnected environments.

  1. Create a filtered RedHat marketplace index image in the private registry

Log into the RedHat registry and the private registry. The private registry is the registry disconnected cluster has access to.

$ docker login registry.redhat.io
$ podman login <private registry>

Prune the index image to include K10 operator(s). The steps below are using the K10 operators from registry.redhat.io/redhat/redhat-marketplace-index:v4.9.

$  opm index prune -f registry.redhat.io/redhat/redhat-marketplace-index:v4.9 \
 -p k10-kasten-operator-rhmp,k10-kasten-operator-term-rhmp \
 -t <private registry>/redhat-marketplace-index:v4.9 \
 -c docker

Push the pruned index image to the private registry

$ docker push <private registry>/redhat-marketplace-index:v4.9

  1. Create a pull secret with RedHat and private registry credentials

Follow the steps in Configuring credentials that allow images to be mirrored to create an image registry credentials file that allows mirroring images to the private registry.

  1. Mirror the operator images to the private registry

$ REG_CREDS=pull-secret.json # path to pull secret file created in step 2

$ oc adm catalog mirror <private registry>/redhat-marketplace-index:v4.9 \
<private registry>/olm-mirror -a ${REG_CREDS}

This copies the operator images from RedHat to the local registry. This also creates a manifest directory which is used in the next two steps.

Example output:

...
info: Mirroring completed in 4.61s (0B/s)
no digest mapping available for <private registry>/redhat-marketplace-index:v4.9, skip writing to ImageContentSourcePolicy
wrote mirroring manifests to <path of manifest directory>

  1. Create an ImageContentSourcePolicy in the disconnected cluster

Create an ImageContentSourcePolicy object using the imageContentSourcePolicy.yaml file in the manifests directory created in step 3.

$ oc create -f <path to manifests dir>/imageContentSourcePolicy.yaml

  1. Create a CatalogSource in the disconnected cluster

Create a CatalogSource object using the catalogSource.yaml file in the manifests directory created in step 3.

$ oc create -f <path to manifests dir>/catalogSource.yaml

catalogSource.yaml can be updated to specify a catalog display name as the example below.

apiVersion: operators.coreos.com/v1alpha1
kind: CatalogSource
metadata:
  name: test-catalog
  namespace: openshift-marketplace
spec:
  image: <catalog image>
  sourceType: grpc
  displayName: Offline Catalog
  publisher: Local Publisher
  updateStrategy:
          registryPoll:
                  interval: 30m

Optionally, default catalog sources can be removed with the command below.

$ oc patch OperatorHub cluster --type json \
  -p '[{"op": "add", "path": "/spec/disableAllDefaultSources", "value": true}]'

Verify the package manifest

$ oc get packagemanifest -n openshift-marketplace
NAME                             CATALOG        AGE
k10-kasten-operator-term-rhmp    Test Catalog   10m
k10-kasten-operator-rhmp         Test Catalog   10m

  1. Install the operators via the operator hub

K10 operators can be now installed from the operator hub.

Follow steps under operator install to continue installing K10.

Other Installation Options

For a complete list of installation options please visit our advanced installation page.

Note

After installing the "Kasten K10 (Enterprise - PAYGO)" edition, if the warning message "Unable to validate Red Hat Marketplace license" is displayed on the K10 dashboard, please verify the cluster is registered with Red Hat Marketplace and the Red Hat Marketplace Operator is installed, and then re-install K10

OpenShift on AWS

When deploying OpenShift on AWS without using the EBS CSI driver for persistent storage, make sure that you configure these policies before executing the installation command provided below:

$ helm install k10 kasten/k10 --namespace=kasten-io \
    --set scc.create=true \
    --set secrets.awsAccessKeyId="${AWS_ACCESS_KEY_ID}" \
    --set secrets.awsSecretAccessKey="${AWS_SECRET_ACCESS_KEY}"

Securing K10 with SecurityContextConstraints

K10 installs customized SecurityContextConstraints (SCC) to ensure that all workloads associated with K10 have just enough privileges to perform their respective tasks.

For additional information about SCCs, please refer to the official OpenShift documentation

SecurityContextConstraints customization

The value of the Priority field in SecurityContextConstraints (SCC) can be adjusted to align the priority with the existing cluster configuration.

To set the desired Priority value in an Operator-managed installation, modify the YAML of the K10 Operand configuration with the parameters below:

scc:
  priority: <priority_value>

This customization can be achieved in a Helm-based installation by adding the following parameter to the Helm command:

--set scc.priority=<priority_value>

SecurityContextConstraints Leakage

Openshift assigns SCC to workloads automatically. By default, the most restrictive SCC matching a workload security requirements will be selected and assigned to that workload. One of the criteria for SCC selection is availability of the SCC to a User or ServiceAccount. SCC leakage means that some workloads might get an SCC applied to them which was not the intended one.

K10 protects its SCC from leaking onto other workloads by limiting access only to its dedicated ServiceAccount:

users:
  - system:serviceaccount:kasten-io:k10-k10

Note

In this example, and in the rest of this page, K10 is installed into the namespace kasten-io (default), the ServiceAccount name is the default one - k10-k10, and the SCC name is also the default one - k10-scc. If the cluster being considered has a different configuration, those values need to be adapted to match the values used during K10's installation in this cluster.

Despite the usage restrictions, it is still possible to get K10's SCC assigned to other workloads. This could happen when a workload is started by a cluster admin or by any other user with allowed use action on all SCCs(*) or on K10's specific SCC(k10-scc). This is because users with the ClusterRole cluster-admin bound to them have unlimited access to all available SCCs, without any restrictions.

K10's SCC may be unexpectedly applied to workloads it wasn't intended for under the following conditions:

  • The workload is initiated by a user with cluster admin privileges

  • The user initiating the workload has a role that grants access to all SCCs

How to verify if access to a specific SecurityContextConstraints is granted

OpenShift's command line (CLI) client, oc, has a can-i command that can be used with impersonation to check if a user can perform a specific action on a specific resource. Alternatively, the standard kubectl CLI client also has the same command built-in and can be used to perform the same check. Simply replace oc by kubectl in the command below.

To check if a user can use/access K10's SCC the following command can be used:

oc auth can-i use securitycontextconstraints/k10-scc --as=<your_username>

The output will contain yes if the specified user is able to use K10's SCC or no if it is not. For example the output for the following check, "Can K10's ServiceAccount use K10's SCC", should be yes:

oc auth can-i use securitycontextconstraints/k10-scc --as=system:serviceaccount:kasten-io:k10-k10

Detailed information about can-i and impersonation can be found in the official Kubernetes documentation.

Accessing Dashboard via Route

As documented here, the K10 dashboard can also be accessed via an OpenShift Route.

Authentication

OpenShift OAuth server

As documented here, the OpenShift OAuth server can be used to authenticate access to K10.

Using OAuth Proxy

As documented here, the OpenShift OAuth proxy can be used for authenticating access to K10.

Validating the Install

To validate that K10 has been installed properly, the following command can be run in K10's namespace (the install default is kasten-io) to watch for the status of all K10 pods:

$ kubectl get pods --namespace kasten-io --watch

It may take a couple of minutes for all pods to come up but all pods should ultimately display the status of Running.

$ kubectl get pods --namespace kasten-io
NAMESPACE     NAME                                    READY   STATUS    RESTARTS   AGE
kasten-io     aggregatedapis-svc-b45d98bb5-w54pr      1/1     Running   0          1m26s
kasten-io     auth-svc-8549fc9c59-9c9fb               1/1     Running   0          1m26s
kasten-io     catalog-svc-f64666fdf-5t5tv             2/2     Running   0          1m26s
...

In the unlikely scenario that pods that are stuck in any other state, please follow the support documentation to debug further.

Validate Dashboard Access

By default, the K10 dashboard will not be exposed externally. To establish a connection to it, use the following kubectl command to forward a local port to the K10 ingress port:

$ kubectl --namespace kasten-io port-forward service/gateway 8080:80

The K10 dashboard will be available at http://127.0.0.1:8080/k10/#/.

For a complete list of options for accessing the Kasten K10 dashboard through a LoadBalancer, Ingress or OpenShift Route you can use the instructions here.