Skip to content


Switch branches/tags

Name already in use

A tag already exists with the provided branch name. Many Git commands accept both tag and branch names, so creating this branch may cause unexpected behavior. Are you sure you want to create this branch?

Latest commit


Git stats


Failed to load latest commit information.
Latest commit message
Commit time

Cass Operator

License: Apache License 2.0

The DataStax Kubernetes Operator for Apache Cassandra®. This repository replaces the old datastax/cass-operator for use-cases in the k8ssandra project. Some documentation is still out of date and will be modified in the future. Check k8ssandra/k8ssandra for more up to date information.

Getting Started

To create a full featured cluster, the recommend approach is to use the Helm charts from k8ssandra. Check the Getting started documentation at ([]. We also provide the ability to install and upgrade the operator with Operator Lifecycle Manager (OLM).

If you wish to use cass-operator only we provide Kustomize templates that can be used to install with kubectl only. This approach is also recommended if you wish to modify the installation beyond our provided examples. If updating from previous versions, please see Upgrade instructions section first.

Supported versions

We actively maintain two different versions of cass-operator, the 1.10.x series as well as the newest version cut from the master branch. The 1.10.x branch receives only bugfixes, while all the new features will only land to the newer versions.

Kubernetes and Openshift version support

1.10.x versions support Kubernetes versions 1.18 to 1.25. For Openshift, 1.10.x supports 4.7 and 4.8 (it will not pass validations in 4.9 due to the use of deprecated APIs).

Newer releases (1.11.0 and up) are tested and supported only in supported Kubernetes versions. That means versions 1.21 and up. For Openshift installations, 1.11.0 and up are supported in 4.9 and newer.

Cassandra version support

For Cassandra versions, 1.10.x will only support 3.11.x and 4.0.x. Support for 4.1 will only land in upcoming versions (1.13.0 or later). Both versions support all DSE 6.8.x releases.

Changelog between versions

The next major/minor version always includes changes from the previously released supported version, but no more than that. That is, if the patch release is released after the next (next in semver) major/minor, the patches are not integrated to it, but are repeated in the changelog for the next version if required (in some cases the bugfixes are not necessary for the next major/minor).

As an example, lets say the changelog has the following release sequence:


1.11.0 will include everything under it, including fixes from 1.10.4, however 1.11.0 does not have the fixes from 1.10.5 (as 1.11.0 was released before 1.10.5). 1.12.0 on the other hand will include all the fixes from 1.10.5, but those fixes are repeated as 1.12.0 lists everything changed from the previous major/minor, that is changes between 1.11.0 and 1.12.0.

While the 1.10.x branch will only have fixes released from that branch, the master branch will include all the versions.

Installing the operator with Kustomize

The operator can be installed in a namespace scoped settings or cluster wide. If installed cluster wide, one can define which namespaces (or all) are watched for CassandraDatacenter objects.

Default installation is simple, the kubectl will create a namespace cass-operator and install cass-operator there. It will only listen for the CassandraDatacenters in that namespace. Note that since the manifests will install a Custom Resource Definition, the user running the commands will need cluster-admin privileges.

Default install requires cert-manager to be installed, since webhooks require TLS certificates to be injected. See below how to install cert-manager if your environment does not have it installed previously.

kubectl apply --force-conflicts --server-side -k

If you wish to install it with cluster wide rights to monitor all the namespaces for CassandraDatacenter objects, use the following command:

kubectl apply --force-conflicts --server-side -k

Alternatively, if you checkout the code, you can use make deploy to run Kustomize and deploy the files.

This will deploy the operator, along with any requisite resources such as Role, RoleBinding, etc., to the cass-operator namespace. You can check to see if the operator is ready as follows:

$ kubectl -n cass-operator get pods --selector name=cass-operator
NAME                             READY   STATUS    RESTARTS   AGE
cass-operator-555577b9f8-zgx6j   1/1     Running   0          25h

Upgrade instructions:

Updates are supported from previous versions of k8ssandra/cass-operator. If upgrading from versions older than 1.7.0 (released under datastax/cass-operator name), please upgrade first to version 1.7.1. The following instructions apply when upgrading from 1.7.1 to 1.8.0 or newer up to 1.10.1. Upgrading to 1.11.0 if using Kubernetes 1.23 requires updating at least to 1.8.0 first, since 1.7.1 can not be used in Kubernetes 1.23 or newer.

If you're upgrading from 1.7.1, there is an additional step to take due to the modifications to cass-operator’s underlying controller-runtime and updated Kubernetes versions. These steps need to be done manually before updating to the newest version of cass-operator. Newer Kubernetes versions require stricter validation and as such we need to remove preserveUnknownFields global property from the CRD to allow us to update to a newer CRD. The newer controller-runtime on the other hand modifies the liveness, readiness and configuration options, which require us to delete the older deployment. These commands do not delete your running Cassandra instances.

Run the following commands assuming cass-operator is installed to cass-operator namespace (change -n parameter if it is installed to some other namespace):

kubectl -n cass-operator delete deployment.apps/cass-operator
kubectl -n cass-operator delete service/cassandradatacenter-webhook-service
kubectl patch crd -p '{"spec":{"preserveUnknownFields":false}}'

You can now install new version of cass-operator as instructed previously.

Install Prometheus monitor rules

If you have Prometheus installed in your cluster, you can apply the following command to install the Prometheus support:

kubectl apply -k

Install cert-manager

We have tested the cass-operator to work with cert-manager versions 1.3.1, 1.5.3 and 1.7.1. Other versions might work also. To install 1.7.1 to your cluster, run the following command:

kubectl apply -f

Modifying the Kustomize template

If you wish to modify the deployment, create your own kustomization.yaml and modify it to your needs. The starting point could be the deployments/config/default and we'll add a cluster scoped installation as our component:

kind: Kustomization

namespace: cass-operator



We provide both components to modify the installation as well as some additional resources for custom features. At the moment, you can modify the behavior of the installation in the following ways, or remove a component to ignore the feature (components enabled in the default installation are marked with asterisk). Apply before component name if using remote installation:

Component name Description
namespace Create namespace before installation*
webhook Enable validation webhooks in cass-operator (requires cert-manager) *
clusterscope Install cass-operator in a cluster scope, monitoring all the namespaces
auth-proxy Protect Prometheus /metrics endpoint with api-server authentication

And following resource. Apply before resource name if using remote installation:

Resource Description
prometheus Add metrics scraping for Prometheus

You can find more resources on how Kustomize works from their documentation. You can install kustomize with make kustomize if you do not have it already (this will install it to bin/kustomize).

Using kustomize to modify the default registry that's used by cass-operator

cass-operator's default image name patterns and repositories are defined in a image_config.yaml file. The image_config.yaml will define only the deployed Cassandra / DSE and server-system-logger images, not the cass-operator itself. To modify imageConfig when deploying, we can create a kustomize component to replace this file in the deployment with our own.

In this example, create two directories, our_installation and private_config_component. You can name these as you wish, just remember to replace the directory names in the yaml files.

In private_config_component create two files, kustomization.yaml and image_config.yaml. Lets say our registry is located at localhost:5000, here's the private_config_component/image_config.yaml:

kind: ImageConfig
  name: image-config
  system-logger: "k8ssandra/system-logger:v1.17.1"
  config-builder: "datastax/cass-config-builder:1.0-ubi7"
  imageRegistry: "localhost:5000"
    repository: "k8ssandra/cass-management-api"
    repository: "datastax/dse-server"
    suffix: "-ubi7"

And for private_config_component/kustomization.yaml you will need the following:

kind: Component

- files:
  - image_config.yaml
  behavior: merge
  name: manager-config

Finally, the kustomization file which we'll deploy will look this (add cluster component if you wish to deploy in cluster-scope), add to our_installation/kustomization.yaml:

kind: Kustomization


  - components/private_image_config

If you also wish to load the cass-operator from a different path, you will need to add the following part to the our_installation/kustomization.yaml:

- name: controller
  newName: localhost:5000/k8ssandra/cass-operator
  newTag: v1.17.1

Run kubectl apply -k our_installation to install cass-operator.

Creating a storage class

If the default StorageClass is not suitable for use (volumeBindingMode WaitForFirstConsumer is required) or you wish to use different one, you will need to create an appropriate storage class which will define the type of storage to use for Cassandra nodes in a cluster. For example, here is a storage class for using SSDs in GKE:

kind: StorageClass
  name: server-storage
  type: pd-ssd
  replication-type: none
volumeBindingMode: WaitForFirstConsumer
reclaimPolicy: Delete

Paste the above to a file and apply:

kubectl apply -f

Creating a CassandraDatacenter

The following resource defines a Cassandra 4.0.1 datacenter with 3 nodes on one rack, which you can also find at config/samples/example-cassdc-three-nodes-single-rack.yaml:

kind: CassandraDatacenter
  name: dc1
  clusterName: development
  serverType: cassandra
  serverVersion: "4.0.1"
    insecure: {}
  size: 3
        storageClassName: server-storage
          - ReadWriteOnce
            storage: 10Gi
      memory: 2Gi
      cpu: 1000m
    securityContext: {}
      - name: cassandra
        securityContext: {}
    - name: rack1
      initial_heap_size: "1G"
      max_heap_size: "1G"
      num_tokens: 16
      authenticator: PasswordAuthenticator
      authorizer: CassandraAuthorizer
      role_manager: CassandraRoleManager

Apply the above as follows:

kubectl -n cass-operator apply -f

You can check the status of pods in the Cassandra cluster as follows:

$ kubectl -n cass-operator get pods --selector
NAME                         READY   STATUS    RESTARTS   AGE
cluster1-dc1-default-sts-0   2/2     Running   0          26h
cluster1-dc1-default-sts-1   2/2     Running   0          26h
cluster1-dc1-default-sts-2   2/2     Running   0          26h

You can check to see the current progress of bringing the Cassandra datacenter online by checking the cassandraOperatorProgress field of the CassandraDatacenter's status sub-resource as follows:

$ kubectl -n cass-operator get cassdc/dc1 -o "jsonpath={.status.cassandraOperatorProgress}"

(cassdc and cassdcs are supported short forms of CassandraDatacenter.)

A value of "Ready", as above, means the operator has finished setting up the Cassandra datacenter.

You can also check the Cassandra cluster status using nodetool by invoking it on one of the pods in the Cluster as follows:

$ kubectl -n cass-operator exec -it -c cassandra cluster1-dc1-default-sts-0 -- nodetool status
Datacenter: dc1
|/ State=Normal/Leaving/Joining/Moving/Stopped
--  Address         Load       Tokens       Owns (effective)  Host ID                               Rack
UN  224.82 KiB  1            65.4%             5e29b4c9-aa69-4d53-97f9-a3e26115e625  r1
UN    186.48 KiB  1            61.6%             b119eae5-2ff4-4b06-b20b-c492474e59a6  r1
UN    205.1 KiB   1            73.1%             0a96e814-dcf6-48b9-a2ca-663686c8a495  r1

The operator creates a secure Cassandra cluster by default, with a new superuser (not the traditional cassandra user) and a random password. You can get those out of a Kubernetes secret and use them to log into your Cassandra cluster for the first time. For example:

$ # get CASS_USER and CASS_PASS variables into the current shell
$ CASS_USER=$(kubectl -n cass-operator get secret cluster1-superuser -o json | jq -r '.data.username' | base64 --decode)
$ CASS_PASS=$(kubectl -n cass-operator get secret cluster1-superuser -o json | jq -r '.data.password' | base64 --decode)
$ kubectl -n cass-operator exec -ti cluster1-dc1-default-sts-0 -c cassandra -- sh -c "cqlsh -u '$CASS_USER' -p '$CASS_PASS'"

Connected to cluster1 at
[cqlsh 5.0.1 | Cassandra 3.11.6 | CQL spec 3.4.4 | Native protocol v4]
Use HELP for help.

cluster1-superuser@cqlsh> select * from system.peers;

 peer      | data_center | host_id                              | preferred_ip | rack    | release_version | rpc_address | schema_version                       | tokens
-----------+-------------+--------------------------------------+--------------+---------+-----------------+-------------+--------------------------------------+-------------------------- |         dc1 | 4bf5e110-6c19-440e-9d97-c013948f007c |         null | default |          3.11.6 | | e84b6a60-24cf-30ca-9b58-452d92911703 | {'-7957039572378599263'} |         dc1 | 3e84b0f1-9c1e-4deb-b6f8-043731eaead4 |         null | default |          3.11.6 | | e84b6a60-24cf-30ca-9b58-452d92911703 | {'-3984092431318102676'}

(2 rows)

Installing cluster via Helm

To install a cluster with optional integrated backup/restore and repair utilities, check the k8ssandra/k8ssandra helm charts project.

If you wish to install only the cass-operator, you can run the following command:

helm repo add k8ssandra
helm install cass-operator k8ssandra/cass-operator -n cass-operator --create-namespace

Installing from OperatorHub / OLM

cass-operator is available in the OperatorHub community under the name of cass-operator-community, but also in the Openshift's certified catalog.


  • Proper token ring initialization, with only one node bootstrapping at a time
  • Seed node management - one per rack, or three per datacenter, whichever is more
  • Server configuration integrated into the CassandraDatacenter CRD
  • Rolling reboot nodes by changing the CRD
  • Store data in a rack-safe way - one replica per cloud AZ
  • Scale up racks evenly with new nodes
  • Scale down racks evenly by decommissioning existing nodes
  • Replace dead/unrecoverable nodes
  • Multi DC clusters (limited to one Kubernetes namespace)

All features are documented in the User Documentation.


The operator is comprised of the following container images working in concert:

The Cassandra container must be built with support for management-api, otherwise cass-operator will fail to work correctly.

Overriding properties of cass-operator created containers

If the CassandraDatacenter specifies a podTemplateSpec field, then containers with specific names can be used to override default settings in containers that will be created by cass-operator.

Currently cass-operator will create an init container with the name of "server-config-init". Containers that will be created have the names "cassandra" and "server-system-logger".

In general, the values specified in this way by the user will override anything generated by cass-operator.

Of special note is that user-specified environment variables, ports, and volumes in the corresponding containers will be added to the values that cass-operator automatically generates for those containers.

kind: CassandraDatacenter
  name: dc1
  clusterName: cluster1
  serverType: cassandra
  serverVersion: 3.11.11
    insecure: {}
  size: 3
        - name: "server-config-init"
          - name: "EXTRA_PARAM"
            value: "123"
        - name: "cassandra"
          terminationMessagePath: "/dev/other-termination-log"
          terminationMessagePolicy: "File"
      storageClassName: server-storage
      - ReadWriteOnce
          storage: 5Gi
      authenticator: org.apache.cassandra.auth.PasswordAuthenticator
      authorizer: org.apache.cassandra.auth.CassandraAuthorizer
      role_manager: org.apache.cassandra.auth.CassandraRoleManager
      initial_heap_size: 800M
      max_heap_size: 800M


  • Kubernetes cluster, 1.21 or newer. For Openshift, version 4.9 or newer. If you're using older versions, please install from 1.10.x branch, which supports Openshift 4.7 and Kubernetes 1.19.


If you wish to file a bug, enhancement proposal or have other questions, use the issues in this repository or in the k8ssandra/k8ssandra repository. PRs should target this repository and you can link the PR to issue repository with k8ssandra/k8ssandra#ticketNumber syntax.

For other means of contacting, check k8ssandra community resources.

Developer setup

Almost every build, test, or development task requires the following pre-requisites...

  • Golang 1.19 or newer
  • Docker, either the packages on Ubuntu, Docker Desktop for Mac, or your preferred docker distribution. Other container engines such as podman should work also.
  • Kind or similar Kubernetes distribution for testing (Docker Desktop / Minikube will work if correct StorageClass is added)


The operator uses Makefiles for its build process.

Build the Operator Container Image

This build task will create the operator container image, building or rebuilding the binary from golang sources if necessary:

make docker-build

Build the Operator Binary

If you wish to perform ONLY to the golang build or rebuild, without creating a container image:

make build

Or just make.

Build and deploy to kind

To simplify testing processes, the following command will build the docker image and load it to the kind instance.

make docker-kind


Tests are separated to unit-tests (including envtests) and end-to-end tests. To run the unit tests and envtests, use:

make test

test target will spawn a envtest environment, which will require the ports to be available. To test the unit test workflow, one can use the act to run the tests with simply running act -j testing. While integration tests will work with act also, we do not recommend that, since it runs them serially and that takes a long time.

End-to-end Automated Testing

Run fully automated end-to-end tests (these will take a while and require a Kubernetes cluster access with up to 6 worker nodes):

make integ-test

To run a single e2e test:

M_INTEG_DIR=test_dir make integ-test

More details of end-to-end testing are here.


This will destroy all of your data!

Delete your CassandraDatacenters first, otherwise Kubernetes will block deletion of the namespace because we use a finalizer. The following command deletes all CassandraDatacenters from all namespaces:

kubectl delete cassdcs --all-namespaces --all

If you used the make deploy to deploy the operator, replace it with make undeploy to uninstall. With the kubectl apply -k option, replace it with kubectl delete -k.


For questions, please reach out on k8ssandra Community channels. For development questions, we are active on our Discord channel #k8ssandra-dev. Or you can open an issue.


Copyright DataStax, Inc.

Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.


For information on the packaged dependencies of Cass Operator and their licenses, check out our open source report.