reshifter

Kubernetes cluster state management

ReShifter is a Kubernetes cluster state management library and tool. You can use it on the command line for one-off tasks, as a Web app with an UI, or in a cron job. It uses the etcd API to query and manipulate the state of Kubernetes-related objects, supporting legacy (v2, /registry), modern (v2/v3, /kubernetes.io) as well as distro-specific keys such as the ones used in the enterprise Kubernetes distro OpenShift.

• Use Cases
• Status and roadmap
• CLI tool
• Web app
  • Deploy app locally
  • Deploy app in OpenShift
  • Deploy app in Kubernetes
• Backup strategies
• Configuration
• Test and development

Use Cases

There are many cases where ReShifter can be useful for you, for example:

• Auditing To comply with regulations, you often have to provide an audit trail. ReShifter allows you to generate one by capturing the state of all objects in a Kubernetes cluster.
• Billing No matter if you charge your (internal) customers by volume or by time, you need to record who has been running what. With ReShifter you can snapshot the state of all objects in a Kubernetes cluster and use this as a basis for billing.
• $aving money When running development or test clusters in a public cloud setting, you might not want to pay for the whole time, only when you actively use it. You can use ReShifter to snapshot the state, shut down the cluster and once ramped up again, restore the state.
• Troubleshooting For any kind of Kubernetes clusters, be it prod or testing, you can use ReShifter to capture the state of the cluster at a certain point in time for offline debugging and troubleshooting or to share the state with a provider, who then can take care of looking into a support case.
• Capacity planning To estimate how many worker nodes you might need to add to your Kubernetes cluster in the future you need to have an idea about the overall load over a longer period of time. ReShifter can help you gathering data points for this capacity planning process.
• Upgrading When you are upgrading a cluster, say, from Kubernetes 1.5 to 1.6, you can use ReShifter to achieve zero-downtime upgrades with minimal manual effort.
• Restore If you are running a Kubernetes cluster in production, you can use ReShifter to back up the current state of the deployed apps. The backup can be useful as a basis for a restore process, into the same or a new cluster.
• Disaster Recovery When operating two or more clusters, you can use ReShifter to replicate the state for DR failover scenarios.

Status and roadmap

See the Trello board for the current status of the work and roadmap items. You can also check out the design philosophy and architecture of ReShifter, if you want to learn more about why it works the way it does.

CLI tool

You can use ReShifter from the CLI via the rcli binaries. Get binary releases for Linux and macOS via the GitHub release page.

$ rcli -h
A CLI for ReShifter, supports backing up and restoring Kubernetes clusters.

Usage:
  rcli [command]

Available Commands:
  backup      Creates a backup of a Kubernetes cluster
  explore     Probes an etcd endpoint
  help        Help about any command
  restore     Performs a restore of a Kubernetes cluster
  stats       Collects stats about Kubernetes-related keys from an etcd endpoint
  version     Displays the ReShifter version

...

Here's a simple usage example which assumes that you've got a Kubernetes cluster with an etcd on http://localhost:4001 running:

# back up Kubernetes cluster to Minio playground:
$ ACCESS_KEY_ID=Q3AM3UQ867SPQQA43P2F \
  SECRET_ACCESS_KEY=zuf+tfteSlswRu7BJ86wekitnifILbZam1KYY3TG \
  rcli backup create --endpoint http://localhost:4001 \
                     --remote play.minio.io:9000 \
                     --bucket mh9-test

# restore cluster from Minio playground, using backup ID 1498815551:
$ ACCESS_KEY_ID=Q3AM3UQ867SPQQA43P2F \
  SECRET_ACCESS_KEY=zuf+tfteSlswRu7BJ86wekitnifILbZam1KYY3TG \
  rcli restore --endpoint http://localhost:4001 \
               --remote play.minio.io:9000 \
               --bucket mh9-test \
               --backupid 1498815551

OK, now that you know about the basic usage, why not learn more by checking out the CLI walkthrough? You can also have a look at the backup strategies as well as the configuration options for more ways to use the ReShifter CLI most effectively.

Web app

If you want to see the app in action, have a look at the app walkthrough.

Deploy app locally

If you want to use the ReShifter app, that is the Web UI, you need to use the Docker image since it bundles the static assets such as HTML, CSS, and JS and the Go binary. For example, to launch the ReShifter app locally, do:

$ docker run --rm -p 8080:8080 quay.io/mhausenblas/reshifter:0.3.21

If you want to use the ReShifter API, for example as a head-less service, you can simply use the binary, no other dependencies required:

$ curl -s -L https://github.com/mhausenblas/reshifter/releases/download/v0.3.21-alpha/reshifter -o reshifter
$ chmod +x reshifter
$ ./reshifter

The ReShifter HTTP API is defined in and available via Swagger: swaggerhub.com/apis/mhausenblas/reshifter/1.0.0

Deploy app in OpenShift

To install the ReShifter app on OpenShift, we use a Makefile that creates a new project called reshifter, launches the app from the Docker image, and finally exposes the service via a route so that you can access the app from outside of the cluster. The following requires access to an OpenShift Origin 1.5 (or OpenShift Container Platform 3.5) cluster and also the oc CLI tool locally installed:

$ make init
$ make build
$ make publish

Deploy app in Kubernetes

To install the ReShifter app in a portable way, use kubectl and the YAML files provided. The following requires access to a vanilla Kubernetes 1.5 or above cluster and also the kubectl CLI tool locally installed:

$ kubectl create -f https://raw.githubusercontent.com/mhausenblas/reshifter/master/deployments/reshifter-app.yaml

Note: the above YAML file defines both a Deployment as well as a Service object for ReShifter. It does, however, not expose the ReShifter service to the outside world. If you want to access the app from outside the cluster, you'll have to use Ingress or adapt the YAML file yourself.

Backup strategies

ReShifter supports different backup strategies.

Configuration

ReShifter takes throughout the deployments (app, HTTP API, CLI) the following configuration parameters, supplied as environment variables. Currently all of them are optional, but may be required depending on the etcd setup in use or what you want to achieve, for example, backing up to an S3-compatible storage.

Note:

• The ACCESS_KEY_ID and SECRET_ACCESS_KEY environment variables are understood as of the AWS guidelines
• The RS_ETCD_* environment variables are understood as of the etcd2 security flags/etcd3 security flags

Test and development

See the pages for testbed and development.