Chapter 2. Using CDK

This section describes how to use the minishift CLI. It covers the basic commands as well as advanced features like profiles, image caching, add-ons and host folders.

For details about using the minishift and oc commands to manage your local OpenShift cluster, see the Interacting with OpenShift section.

2.1. Basic Usage

2.1.1. Overview

When you use CDK, you interact with the following components:

  • the CDK virtual machine (VM)
  • the Docker daemon running on the VM
  • the OpenShift cluster running on the Docker daemon

The CDK architecture diagram outlines these components. The minishift binary, placed on the PATH for easy execution, is used to start, stop, and delete the CDK VM. The VM itself is bootstrapped off of a Red Hat Enterprise Linux Live ISO.

Some CDK commands, for example docker-env, interact with the Docker daemon, whilst others communicate with the OpenShift cluster, for example the openshift command.

Once the OpenShift cluster is up and running, you interact with it using the oc binary. CDK caches this binary under $MINISHIFT_HOME (per default ~/.minishift). minishift oc-env is an easy way to add the oc binary to your PATH.

For more details about using CDK to manage your local OpenShift cluster, see the Interacting with OpenShift section.

Figure 2.1. : Minishift architecture

: Minishift architecture

2.1.2. CDK Life-cycle The minishift setup-cdk Command

The minishift setup-cdk command copies a Red Hat Enterprise Linux ISO image to the CDK cache, creates a default configuration file, installs and enables default add-ons, and creates a marker file to confirm that the command has been run.

The command also copies the oc binary to your host so that you can interact with OpenShift through the oc command line tool or through the Web console, which can be accessed through the URL provided in the output of the minishift start command. The minishift start Command

The minishift start command creates and configures the CDK VM and provisions a local, single-node OpenShift cluster within the CDK VM. The minishift stop Command

The minishift stop command stops your OpenShift cluster and shuts down the CDK VM, but preserves the OpenShift cluster state.

Starting CDK again will restore the OpenShift cluster, allowing you to continue working from the last session. However, you must enter the same parameters that you used in the original start command.

Efforts to further refine this experience are in progress. For details, see the GitHub issue #179. The minishift delete Command

The minishift delete command deletes the OpenShift cluster, and also shuts down and deletes the CDK VM. No data or state are preserved.

2.1.3. Runtime Options

The runtime behavior of CDK can be controlled through flags, environment variables, and persistent configuration options.

The following precedence order is applied to control the behavior of CDK. Each action in the following list takes precedence over the action below it:

  1. Use command line flags as specified in the Flags section.
  2. Set environment variables as described in the Environment Variables section.
  3. Use persistent configuration options as described in the Persistent Configuration section.
  4. Accept the default value as defined by CDK. Flags

You can use command line flags with CDK to specify options and direct its behavior. This has the highest precedence. Almost all commands have flags, although different commands might have different flags. Some of the commonly-used command line flags of the minishift start command are cpus, memory or vm-driver. Environment Variables

CDK allows you to specify command line flags you commonly use through environment variables. To do so, apply the following rules to the flag you want to set as an environment variable.

  1. Apply MINISHIFT_ as a prefix to the flag you want to set as an environment variable. For example, the vm-driver flag of the minishift start command becomes MINISHIFT_vm-driver.
  2. Use uppercase characters for the flag, so MINISHIFT_vm-driver in the above example becomes MINISHIFT_VM-DRIVER.
  3. Replace - with _, so MINISHIFT_VM-DRIVER becomes MINISHIFT_VM_DRIVER.

Environment variables can be used to replace any option of any CDK command. A common example is the URL of the ISO to be used. Usually, you specify it with the iso-url flag of the minishift start command. Applying the above rules, you can also specify this URL by setting the environment variable as MINISHIFT_ISO_URL.


You can also use the MINISHIFT_HOME environment variable, to choose a different home directory for CDK. By default, the minishift command places all runtime state into ~/.minishift. This environment variable is currently experimental and semantics might change in future releases.

To use MINISHIFT_HOME, you should set the new home directory when you first set up CDK. For example, this sets the minishift home directory to ~/.mynewdir on a Linux system:

$ minishift setup-cdk --minishift-home ~/.mynewdir
$ export MINISHIFT_HOME=~/.mynewdir
$ echo 'export MINISHIFT_HOME=~/.mynewdir' >> ~/.bashrc Persistent Configuration

Using persistent configuration allows you to control CDK behavior without specifying actual command line flags, similar to the way you use environment variables.

CDK maintains a configuration file in $MINISHIFT_HOME/config/config.json. This file can be used to set commonly-used command line flags persistently.


Persistent configuration can only be applied to the set of supported configuration options that are listed in the synopsis of the minishift config sub-command. This is different from environment variables, which can be used to replace any option of any command. Setting Persistent Configuration Values

The easiest way to change a persistent configuration option is with the minishift config set sub-command. For example:

# Set default memory 4096 MB
$ minishift config set memory 4096

Flags which can be used multiple times per command invocation, like docker-env or insecure-registry, need to be comma-separated when used with the config set command. For example, from the CLI, you can use insecure-registry like this:

$ minishift start --insecure-registry --insecure-registry

To configure the same registries in the persistent configuration, run the following command:

$ minishift config set insecure-registry,

To view all persistent configuration values, you can use the minishift config view sub-command:

$ minishift config view
- memory: 4096

Alternatively, you can display a single value with the minishift config get sub-command:

$ minishift config get memory
4096 Unsetting Persistent Configuration Values

To remove a persistent configuration option, you can use the minishift config unset sub-command. For example:

$ minishift config unset memory Driver-Specific Environment Variables

You can also set driver-specific environment variables. Each Docker Machine driver supports its own set of options and variables. A good starting point is the official Docker Machine driver documentation.

xhyve and KVM documentation is available under their respective GitHub repositories (docker-machine-driver-xhyve and docker-machine-kvm).

To use driver-specific options, make sure to export the variable as defined in its driver documentation before running minishift start. For example, the xhyve experimental NFS sharing can be enabled by executing:

$ minishift start --vm-driver xhyve

Driver-specific options might overlap with values specified using CDK-specific flags and environment variables. Examples are boot2docker URL, memory size, cpu count, and so on. In this case, driver-specific environment variables will override CDK-specific settings.

2.1.4. Persistent Volumes

As part of the OpenShift cluster provisioning, 100 persistent volumes are created for your OpenShift cluster. This allows applications to make persistent volumes claims. The location of the persistent data is determined in the host-pv-dir flag of the minishift start command and defaults to /var/lib/minishift/openshift.local.pv on the CDK VM.

2.1.5. HTTP/HTTPS Proxies

If you are behind an HTTP/HTTPS proxy, you need to supply proxy options to allow Docker and OpenShift to work properly. To do this, pass the required flags during minishift start.

For example:

$ minishift start --http-proxy http://YOURPROXY:PORT --https-proxy https://YOURPROXY:PORT

In an authenticated proxy environment, the proxy_user and proxy_password must be a part of proxy URI.

 $ minishift start --http-proxy http://<proxy_username>:<proxy_password>@YOURPROXY:PORT \
                   --https-proxy https://<proxy_username>:<proxy_password>@YOURPROXY:PORT

You can also use the --no-proxy flag to specify a comma-separated list of hosts that should not be proxied.

Using the proxy options will transparently configure the Docker daemon as well as OpenShift to use the specified proxies.

  • minishift start honors the environment variables HTTP_PROXY, HTTPS_PROXY and NO_PROXY. If these variables are set, they are implicitly used during minishift start unless explicitly overridden by the corresponding command line flags.
  • Use the minishift start --ocp-tag flag to request a specific version of OpenShift Container Platform. You can list all CDK-compatible OpenShift Container Platform versions with the minishift openshift version list command.

2.1.6. Networking

The CDK VM is exposed to the host system with a host-only IP address that can be obtained with the minishift ip command.

2.1.7. Connecting to the CDK VM with SSH

You can use the minishift ssh command to interact with the CDK VM.

You can run minishift ssh without a sub-command to open an interactive shell and run commands on the CDK VM in the same way that you run commands interactively on any remote machine using SSH.

You can also run minishift ssh with a sub-command to send the sub-command directly to the CDK VM and return the result to your local shell. For example:

$ minishift ssh -- docker ps
CONTAINER    IMAGE                   COMMAND                CREATED        STATUS        NAMES
71fe8ff16548 openshift3/ose:v3.9.14  "/usr/bin/openshift s" 4 minutes ago  Up 4 minutes  origin

2.2. CDK Profiles

2.2.1. Overview


You must run minishift setup-cdk before using profiles.

A profile is an instance of the Minishift VM along with all of its configuration and state. The profile feature allows you to create and manage these isolated instances of Minishift.

Each CDK profile is created with its own configuration (memory, CPU, disk size, add-ons, and so on) and is independent of other profiles. Refer to the use of environment variables if you want to make sure that certain configuration settings, for example cpus or memory, get applied to all profiles.

The active profile is the profile against which all commands are executed, unless the global --profile flag is used. You can determine the active profile by using the minishift profile list command. You can execute single commands against a non-active profile by using the --profile flag, for example minishift --profile profile-demo console to open the OpenShift console for the specified profile-demo profile.

On top of the --profile flag, there are commands for listing, deleting and setting the active profile. These commands are described in the following sections.


Even though profiles are independent of each other, they share the same cache for ISOs, oc binaries and container images. minishift delete --clear-cache will affect all profiles for this reason. We recommend using --clear-cache with caution.

2.2.2. Creating Profiles

There are two ways to create a new profile.


Profile names can only consist of alphanumeric characters. Underscores ( _ ) and hyphens ( - ) are allowed as separators. Using the --profile Flag

When you run the minishift start command with the --profile flag the profile is created if it does not already exist. For example:

$ minishift --profile profile-demo start
-- Checking if requested hypervisor 'xhyve' is supported on this platform ... OK
-- Checking if xhyve driver is installed ...
   Driver is available at /usr/local/bin/docker-machine-driver-xhyve
   Checking for setuid bit ... OK
-- Checking the ISO URL ... OK
-- Starting local OpenShift cluster using 'xhyve' hypervisor ...
-- Minishift VM will be configured with ...
   Memory:    2 GB
   vCPUs :    2
   Disk size: 20 GB

See also Example Workflow for Profile Configuration.


A profile automatically becomes the active profile when a CDK instance is started successfully via minishift start. Using the profile set Command

The other option to create a profile is to use the profile set command. If the specified profile does not exist, it is implicitly created:

$ minishift profile set demo
Profile 'demo' set as active profile

The default profile is minishift. It will be present by default and it does not need to be created.

2.2.3. Listing Profiles

You can list all existing profiles with the minishift profile list command. You can also see the active profile highlighted in the output.

$ minishift profile list
- minishift     Running     	(Active)
- profile-demo  Does Not Exist

2.2.4. Switching Profiles

To switch between profiles use the minishift profile set command:

$ minishift profile set profile-demo
Profile 'profile-demo' set as active profile

Only one profile can be active at any time.

2.2.5. Deleting Profiles

To delete a profile, run:

$ minishift profile delete profile-demo
You are deleting the active profile. It will remove the VM and all related artifacts. Do you want to continue [y/N]?: y
Deleted:  /Users/john/.minishift/profiles/profile-demo
Profile 'profile-demo' deleted successfully
Switching to default profile 'minishift' as the active profile.

The default profile minishift cannot be deleted.

2.2.6. Example Workflow for Profile Configuration

You have two options to create a new profile and configure its persistent configuration. The first option is to implicitly create the new profile by making it the active profile using the profile set command. Once the profile is active you can run any minishift config command. Lastly, start the instance:

$ minishift profile set profile-demo
$ minishift config set memory 8GB
$ minishift config set cpus 4
$ minishift addon enable anyuid
$ minishift start

The alternative is to execute a series of commands each specifying the targeted profile explicitly using the --profile flag:

$ minishift --profile profile-demo config set memory 8GB
$ minishift --profile profile-demo config set cpus 4
$ minishift --profile profile-demo addon enable anyuid
$ minishift --profile profile-demo minishift start

2.3. Image Caching

2.3.1. Overview

To speed up the provisioning of the OpenShift cluster and to minimize network traffic, container images can be cached on the host. These images can then be imported into the running Docker daemon, either explicitly on request or implicitly during minishift start. The following sections describe image caching and its configuration in more detail.


The format in which images are cached has changed with CDK version 3.3.0. Prior to 3.3.0, the images were stored as tar files. As of 3.3.0, images are stored in the OCI image format.

If you used image caching prior to CDK 3.3.0, your cache will need to be recreated. If you want to remove the obsolete pre-3.3.0 images, you can clear your cache via:

$ minishift delete --clear-cache

2.3.2. Explicit Image Caching

CDK provides the image command together with its sub-commands to control the behavior of image caching. To export and import images from the Docker daemon of the CDK VM, use minishift image export and minishift image import. Importing and Exporting Single Images

Once the CDK VM is running, images can be explicitly exported from the Docker daemon:

$ minishift image export <image-name-0> <image-name-1> ...
Pulling image <image-name-0> .. OK
Exporting <image-name-0>. OK
Pulling image <image-name-1> .. OK
Exporting <image-name-2>. OK

Images which are not available in the Docker daemon will be pulled prior to being exported to the host.

To import previously cached images, use the minishift image import command:

$ minishift image import <image-name-0> <image-name-1> ...
Importing <image-name-0> . OK Listing Cached Images

The minishift image list command lists either the currently cached images or the images available in the CDK Docker daemon.

To view currently cached images on the host:

$ minishift image list

To view images available in the Docker daemon:

$ minishift image list --vm
openshift/origin:v3.6.0 Persisting Cached Image Names

In order to avoid having to type the image names explicitly as part of the image export or image import command, you can store a list of image names for import and export in the persistent configuration.

Use minishift image cache-config view to view the list of currently configured images and minishift image cache-config add to add images to the list:

$ minishift image cache-config view
$ minishift image cache-config add alpine:latest busybox:latest
$ minishift image cache-config view

To remove images from the list use minishift image cache-config remove:

$ minishift image cache-config remove alpine:latest
$ minishift image cache-config view

Once the image names are stored in the persistent configuration, you can run minishift image export and minishift image import without any arguments. Exporting and Importing All Images

You can export and import all images using the --all flag. For the export command, this means that all images currently available on the Docker daemon will be exported to the host. For the import command, it means that all images available in the local CDK cache will be imported into the Docker daemon of the CDK VM.


Exporting and importing all images can take a long time and locally cached images can take up a considerable amount of disk space. We recommend using this feature with caution.

2.3.3. Implicit Image Caching

Image caching is enabled by default for CDK. It occurs in a background process after the minishift start command is completed for the first time. Once the images are cached under $MINISHIFT_HOME/cache/image, successive CDK VM creations will use these cached images.

To disable this feature you need to disable the image-caching property in the persistent configuration using the minishift config set command:

$ minishift config set image-caching false

Implicit image caching will transparently add the required OpenShift images to the list of cached images as specified per cache-images configuration option. See Persisting Cached Image Names.


Each time an image exporting background process runs, a log file is generated under $MINISHIFT_HOME/logs which can be used to verify the progress of the export.

You can re-enable the caching of the OpenShift images by setting image-caching to true or removing the setting altogether using minishift config unset:

$ minishift config unset image-caching

2.4. Add-ons

2.4.1. Overview

CDK allows you to extend the vanilla OpenShift setup provided by cluster up with an add-on mechanism.

Add-ons are directories that contain a text file with the .addon extension. The directory can also contain other resource files such as JSON template files. However, only one .addon file is allowed per add-on.

The following example shows the contents of an add-on file, including the name and description of the add-on, additional metadata, and the actual add-on commands to apply.

Example: anyuid add-on definition file

# Name: anyuid                                                                          1
# Description: Allows authenticated users to run images under a non pre-allocated UID   2
# Required-Vars: ACME_TOKEN                                                             3
# OpenShift-Version: >3.6.0                                                             4

oc adm policy add-scc-to-group anyuid system:authenticated                              5

(Required) Name of the add-on.
(Required) Description of the add-on.
(Optional) Comma separated list of required interpolation variables. See Variable Interpolation.
(Optional) OpenShift version required to run a specific add-on. See OpenShift Version Semantics.
Actual add-on command. In this case, the command executes the oc binary.
  • Comment lines, starting with the '#' character, can be inserted anywhere in the file.
  • Commands starting with the ! character ignore execution failure. With the help of this, add-on will be idempotent i.e. a command can be executed multiple times without changing the final behavior of the add-on.

Enabled add-ons are applied during minishift start, immediately after the initial cluster provisioning is successfully completed.

2.4.2. OpenShift-Version Semantics

As part of the add-on metadata you can specify the OpenShift version which needs to be running in order to apply the add-on. To do so, you can specify the optional OpenShift-Version metadata field. The semantics are as follows:


OpenShift version higher than 3.6.0 needs to be running in order to apply the add-on.


OpenShift version 3.6.0 or higher needs to be running in order to apply the add-on.


OpenShift version 3.6.0 needs to be running in order to apply the add-on.

>=3.5.0, <3.8.0

OpenShift version should higher or equal to 3.5.0 but lower than 3.8.0.

>=3.5.0, <=3.8.0

OpenShift version should higher or equal to 3.5.0 but lower or equal to 3.8.0.

  • If the metadata field OpenShift-Version is not specified in the add-on header, the add-on can be applied against any version of OpenShift.
  • OpenShift-Version only supports versions in the form of <major>.<minor>.<patch>.

2.4.3. Add-on Commands

This section describes the commands that an add-on file can contain. They form a small domain-specific language for CDK add-ons:

If the add-on command starts with ssh, you can run any command within the CDK-managed VM. This is similar to running minishift ssh and then executing any command on the VM. For more information about minishift ssh command usage, see Connecting to the CDK VM with SSH.

If the add-on command starts with oc, it uses the oc binary that is cached on your host to execute the specified oc command. This is similar to running oc --as system:admin …​ from the command line.


The oc command is executed as system:admin.

If the add-on command starts with openshift, you can run the openshift binary within the container that runs OpenShift. This means that any file parameters or other system-specific parameters must match the environment of the container instead of your host.
If the add-on command starts with docker, it executes a docker command against the Docker daemon within the CDK VM. This is the same daemon on which the single-node OpenShift cluster is running as well. This is similar to running eval $(minishift docker-env) on your host and then executing any docker command. See also minishift docker-env.
If the add-on command starts with echo, the arguments following the echo command are printed to the console. This can be used to provide additional feedback during add-on execution.
If the add-on command starts with sleep, it waits for the specified number of seconds. This can be useful in cases where you know that a command such as oc might take a few seconds before a certain resource can be queried.

Trying to use an undefined command will cause an error when the add-on gets parsed.

2.4.4. Variable Interpolation

CDK allows the use of variables within the add-on commands. Variables have the format #{<variable-name>}. The following example shows how the OpenShift routing suffix can be interpolated into an openshift command to create a new certificate as part of securing the OpenShift registry. The used variable #{routing-suffix} is part of the built-in add-on variables.

Example: Usage of the routing-suffix variable

$ openshift admin ca create-server-cert \
  --signer-cert=/var/lib/origin/openshift.local.config/master/ca.crt \
  --signer-key=/var/lib/origin/openshift.local.config/master/ca.key \
  --signer-serial=/var/lib/origin/openshift.local.config/master/ca.serial.txt \
  --hostnames='docker-registry-default.#{routing-suffix},docker-registry.default.svc.cluster.local,' \
  --cert=/etc/secrets/registry.crt \
  --key=/etc/secrets/registry.key Built-in Variables

There exist several built-in variables which are available for interpolation at all times. The following table shows these variables.

Table 2.1. Supported built-in add-on variables



IP of the CDK VM.


OpenShift routing suffix for the application.


Name of the current add-on. Dynamic Variables

The commands minishift start as well as minishift addons apply also provide an --addon-env flag which allows to dynamically pass variables for interpolation, for example:

$ minishift addons apply --addon-env PROJECT_USER=john acme

The --addon-env flag can be specified multiple times to define more than one variables for interpolation.

Specifying dynamic variables also works in conjunction with setting persistent configuration values.

$ minishift config set addon-env PROJECT_USER=john
$ minishift addons apply acme

Multiple variables must be comma separated when the minishift config set command is used.

There is also the possibility to dynamically interpolate a variable with the value of an environment variable at the time the add-on gets applied. For this the variable value needs to be prefixed with env.

$ minishift config set addon-env PROJECT_USER=env.USER        1
$ minishift addons apply acme                                 2
Using the env prefix ensures that instead of literally replacing '#{PROJECT_USER}' with 'env.USER', the value of the environment variable USER is used. If the environment variable is not set, interpolation does not occur.
When the add-on is applied, each occurrence of #{PROJECT_USER} within an add-on command gets replaced with the value of the environment variable USER.

As an add-on developer, you can enforce that a variable value is provided when the add-on gets applied by adding the variable name to the Required-Vars metadata header. Multiple variables need to be comma separated.

# Name: acme
# Description: ACME add-on
# Required-Vars: PROJECT_USER

You can also provide default values for variables using the Var-Defaults metadata header. Var-Defaults needs to be specified in the format of <key>=<value>. Multiple default key/value pairs need to be comma separated.

# Name: acme
# Description: ACME add-on
# Required-Vars: PROJECT_USER
# Var-Defaults: PROJECT_USER=john
  • = and , are metacharacters and cannot be used as part of keys or values.
  • Variable values specified via the command line using the --addon-env or set via minishift config set addon-env have precedence over Var-Defaults.

2.4.5. Default Add-ons

CDK provides a set of built-in add-ons that offer some common OpenShift customization to assist with development. During minishift setup-cdk, Minishift automatically installs and enables the xpaas, anyuid, and admin-user add-ons. To install the default add-ons, run:

$ minishift addons install --defaults

This command extracts the default add-ons to the add-on installation directory $MINISHIFT_HOME/addons. To view the list of installed add-ons, you can then run:

$ minishift addons list --verbose=true

This command prints a list of installed add-ons. You should at least see the anyuid add-on listed. This is an important add-on that allows you to run images that do not use a pre-allocated UID. By default, this is not allowed in OpenShift.

Table 2.2. Default add-ons

Add-on NameDescription


Changes the default security context constraints to allow pods to run with arbitrary UID.


Creates a user named 'admin' and assigns the cluster-admin role to it.


Runs an instance of the Eclipse Che integrated development environment.


Creates an edge terminated route for the OpenShift registry.


Imports xPaaS templates.


The 'che' add-on is provided as a Technology Preview. Support for this add-on falls under the Technology Preview Features Support Scope. Add-ons by the Community

Apart from the several default add-ons, there are a number of community developed add-ons for CDK. Community add-ons can be found in the minishift-addons repository. You can get all the information about these add-ons in the repository. The instructions for installing them can be found in the README.

2.4.6. Installing Add-ons

Add-ons are installed with the minishift addons install command.

The following example shows how to install an add-on.

Example: Installing an add-on

$ minishift addons install <path_to_addon_directory>

2.4.7. Enabling and Disabling Add-ons

Add-ons are enabled with the minishift addons enable command and disabled with the minishift addons disable command. Enabled add-ons automatically get executed during minishift start.

The following examples show how to enable and disable the anyuid add-on.

Example: Enabling the anyuid add-on

$ minishift addons enable anyuid

Example: Disabling the anyuid add-on

$ minishift addons disable anyuid Add-on Priorities

When you enable an add-on, you can also specify a priority, which determines the order that the add-ons are applied.

The following example shows how to enable the registry add-on with a higher priority value.

Example: Enabling the registry add-on with priority

$ minishift addons enable registry --priority=5

The add-on priority attribute determines the order in which add-ons are applied. By default, an add-on has the priority 0. Add-ons with a lower priority value are applied first.

In the following example, the anyuid, registry, and eap add-ons are enabled with the respective priorities of 0, 5 and 10. This means that anyuid is applied first, followed by registry, and lastly the eap add-on.

Example: List command output with explicit priorities

$ minishift addons list
- anyuid         : enabled    P(0)
- registry       : enabled    P(5)
- eap            : enabled    P(10)


If two add-ons have the same priority the order in which they are getting applied is not determined.

2.4.8. Applying Add-ons

Add-ons can be explicitly executed with the minishift addons apply command. You can use the apply command for both enabled and disabled add-ons. To apply multiple add-ons with a single command, specify add-on names separated by space.

The following example shows how to explicitly apply the anyuid and the admin-user add-ons.

Example: Applying anyuid and admin-user add-ons

$ minishift addons apply anyuid admin-user

2.4.9. Removing Add-ons

Add-ons can be removed with the minishift addons remove command. It is the mirror command to minishift addons apply and similarly can be used regardless of whether the add-on is enabled or not. Provided the specified add-on is installed and has a <addon_name>.addon.remove file, minishift addons remove will execute the commands specified in this file.

To remove multiple add-ons with a single command, specify the add-on names separated by a space. The following example shows how to explicitly remove the admin-user add-on.

Example: Removing admin-user add-on

$ minishift addons remove admin-user
-- Removing addon 'admin-user':.
admin user deleted

2.4.10. Uninstalling Add-ons

Add-ons can be uninstalled with the minishift addons uninstall command. It is the mirror command to minishift addons install and can be used regardless of whether the add-on is enabled or not. Provided the specified add-on is installed, minishift addons uninstall will delete the corresponding add-on directory from $MINISHIFT_HOME/addons.

The following example shows how to explicitly uninstall the admin-user add-on:

Example: Uninstalling admin-user add-on

$ minishift addons uninstall admin-user
Add-on 'admin-user' uninstalled

2.4.11. Writing Custom Add-ons

To write a custom add-on, you should create a directory and in it create at least one text file with the extension .addon, for example admin-role.addon.

This file needs to contain the Name and Description metadata fields, as well as the commands that you want to execute as a part of the add-on.

The following example shows the definition of an add-on that gives the developer user cluster-admin privileges.

Example: Add-on definition for admin-role

# Name: admin-role
# Description: Gives the developer user cluster-admin privileges

oc adm policy add-role-to-user cluster-admin developer

After you define the add-on, you can install it by running:

$ minishift addons install <ADDON_DIR_PATH>

You can also write metadata with multiple lines.

Example: Add-on definition which contain multiline description

# Name: prometheus
# Description: This template creates a Prometheus instance preconfigured to gather OpenShift and
# Kubernetes platform and node metrics and report them to admins. It is protected by an
# OAuth proxy that only allows access for users who have view access to the prometheus
# namespace. You may customize where the images (built from openshift/prometheus
# and openshift/oauth-proxy) are pulled from via template parameters.
# Url:

You can also edit your add-on directly in the CDK add-on install directory $MINISHIFT_HOME/addons. Be aware that if there is an error in the add-on, it will not show when you run any addons commands, and it will not be applied during the minishift start process.

To provide add-on removal instructions, you can create text file with the extension .addon.remove, for example admin-user.addon.remove. Similar to the .addon file, it needs the Name and Description metadata fields. If a .addon.remove file exists, it can be applied via the remove command.

2.5. Host Folders

2.5.1. Overview

Host folders are directories on the host which are shared between the host and the CDK VM. This allows for two way file synchronization between the host and the VM. The following sections discuss usage of the minishift hostfolder command.

2.5.2. The minishift hostfolder Command

CDK provides the minishift hostfolder command to list, add, mount, unmount and remove host folders. You can use the hostfolder command to mount multiple shared folders onto custom specified mount points.

  • Currently CIFS and SSHFS based host folders are supported. SSHFS support is experimental. Prerequisites CIFS

To use the minishift hostfolder command for CIFS based host folders, you need to be able to share directories using CIFS. On Windows, CIFS is the default technology for sharing directories. For example, on Windows 10 the C:\Users directory is shared by default and can be accessed by locally-authenticated users.

It is also possible to use CIFS on macOS and Linux. On macOS, you can enable CIFS-based shares under System Preferences > Sharing. See How to connect with File Sharing on your Mac for detailed setup instructions.

On Linux, follow your distribution-specific instructions to install Samba. Refer to Samba File and Print Server in RHEL to learn how to configure the Samba implementation of CIFS in Red Hat Enterprise Linux. SSHFS

SSHFS based host folders work without prerequisites. Displaying Host Folders

The minishift hostfolder list command gives you an overview of the defined host folders, their names, mount points, remote paths and whether they are currently mounted.

An example output could look like:

$ minishift hostfolder list
Name    Type    Source               Mountpoint       Mounted
test    sshfs   /Users/john/test     /mnt/sda1/test   N

In this example, there is an SSHFS based host folder with the name test which mounts /Users/john/test onto /mnt/sda1/test in the CDK VM. The share is currently not mounted. Adding Host Folders

The minishift hostfolder add command allows you to define a new host folder.

The exact syntax to use depends on the host folder type. Independent of the type you can choose between non-interactive and interactive configuration. The default is non-interactive. By specifying the --interactive flag you can select the interactive configuration mode.

The following sections give examples for configuring CIFS and SSHFS host folders. CIFS

Adding a CIFS based hostfolder

$ minishift hostfolder add -t cifs --source // --target /mnt/sda1/myshare --options username=john,password=mysecret,domain=MYDOMAIN myshare

The above command will create a new host folder with the name of myshare. The source of the host folder is the UNC path // which is required. The target or mount point is /mnt/sda1/myshare within the CDK VM. Using the --options flag host folder type specific options can be specified using a list of comma separated key/value pairs. In the case of a CIFS host folder the required options are the user name for the CIFS share as well as the password. Via the the --options flag the domain of the share can be specified as well. Often this can be left out, but for example on Windows, when your account is linked to a Microsoft account, you must use the Microsoft account email address as user name as well as your machine name as displayed by $env:COMPUTERNAME as a domain.


On Windows hosts, the minishift hostfolder add command also provides a users-share option. When this option is specified, no UNC path needs to be specified and C:\Users is assumed. SSHFS

Adding an SSHFS based hostfolder

$ minishift hostfolder add -t sshfs --source /Users/john/myshare --target /mnt/sda1/myshare myshare

For the case of an SSHFS based host folder only the source and target of the host folder need to be specified. Instance-Specific Host Folders

By default, host folder definitions are persistent, similar to other persistent configuration options. This means that these host folder definitions will survive the deletion and subsequent re-creation of a CDK VM.

In some cases you might want to define a host folder just for a specific CDK instance. To do so, you can use the --instance-only flag of the minishift hostfolder add command. Host folder definitions that are created with the --instance-only flag will be removed together with any other instance-specific state during minishift delete. Mounting Host Folders

After you have added host folders, you use the minishift hostfolder mount command to mount a host folder by its name:

$ minishift hostfolder mount myshare

You can verify that the host folder is mounted by running:

$ minishift hostfolder list
Name       Mountpoint          Remote path              Mounted
myshare    /mnt/sda1/myshare   //   Y

Alternatively, you can list the actual content of the mounted host folder:

$ minishift ssh "ls -al /mnt/sda1/myshare"

When mounting SSHFS based host folders an SFTP server process is started on port 2022 of the host. Make sure that your network and firewall settings allow this port to be opened. If you need to configure this post you can make use of CDK’s persistent configuration using the key hostfolders-sftp-port, for example:

$ minishift config set hostfolders-sftp-port 2222 Auto-Mounting Host Folders

Host folders can also be mounted automatically each time you run minishift start. To set auto-mounting, you need to set the hostfolders-automount option in the CDK configuration file.

$ minishift config set hostfolders-automount true

After the hostfolders-automount option is set, CDK will attempt to mount all defined host folders during minishift start. Unmounting Host Folders

You use the minishift hostfolder umount command to unmount a host folder.

$ minishift hostfolder umount myshare

$ minishift hostfolder list
Name       Mountpoint          Remote path              Mounted
myshare    /mnt/sda1/myshare   //   N Deleting Host Folders

You use the minishift hostfolder remove command to remove a host folder definition.

$ minishift hostfolder list
Name        Mountpoint            Remote path              Mounted
myshare     /mnt/sda1/myshare     //   N

$ minishift hostfolder remove myshare

$ minishift hostfolder list
No host folders defined

2.6. Assign Static IP Address

2.6.1. Overview

Most hypervisors do not support extending the lease time when the IP is assigned using DHCP. This might lead to a new IP being assigned to the VM after a restart as it will conflict with the security certificates generated for the old IP. This will make CDK completely unusable until a new instance is set up by running minishift delete followed by minishift start.

To prevent this, CDK includes the functionality to set a static IP address to the VM. This will prevent the IP address from changing between restarts. However, it will not work on all of the driver plug-ins at the moment due to the way the IP address is resolved.

  • Assigning a static IP address to the CDK VM is only officially supported for Hyper-V.
  • The CDK VM cannot be assigned a static IP address when using the KVM driver plug-in.
  • An experimental --set-static flag is available for the minishift ip command for other hypervisors. See Set Fixed IP Address for more information.

2.6.2. Assign IP Address to Hyper-V

Since the Internal Virtual Switch for Hyper-V does not provide a DHCP offer option, an IP address needs to be provided in a different way. Functionality is provided to assign an IP address on startup using the Data Exchange Service for Hyper-V.

To make this work, you need to create a Virtual Switch using NAT.


WinNAT is limited to one NAT network per host. For more details about capabilities and limitations, please see the WinNAT capabilities and limitations blog.

The following command will attempt to assign an IP address for use on the Internal Virtual Switch 'MyInternal':

PS> minishift.exe start `
  --network-ipaddress `
  --network-gateway `

If you want to use the 'DockerNAT' network, the following commands are needed to setup the correct NAT networking and assign an IP in the range expected:

PS> New-NetNat -Name SharedNAT -InternalIPInterfaceAddressPrefix
PS> minishift.exe start `
  --network-ipaddress `
  --network-gateway `

Be sure to specify a valid gateway and nameserver. Failing to do so will result in connectivity issues.

2.7. CDK Docker daemon

2.7.1. Overview

When running OpenShift in a single VM, you can reuse the Docker daemon managed by CDK for other Docker use-cases as well. By using the same Docker daemon as CDK, you can speed up your local development.

2.7.2. Console Configuration

In order to configure your console to reuse the CDK Docker daemon, follow these steps:

  1. Make sure that you have the Docker client binary installed on your machine. For information about specific binary installations for your operating system, see the Docker installation site.
  2. Start CDK with the minishift start command.
  3. Run the minishift docker-env command to display the command you need to type into your shell in order to configure your Docker client. The command output will differ depending on OS and shell type.

    $ minishift docker-env
    export DOCKER_TLS_VERIFY="1"
    export DOCKER_HOST="tcp://"
    export DOCKER_CERT_PATH="/Users/john/.minishift/certs"
    export DOCKER_API_VERSION="1.24"
    # Run this command to configure your shell:
    # eval $(minishift docker-env)
  4. Test the connection by running the following command:

    $ docker ps

    If successful, the shell will print a list of running containers.

2.8. Experimental Features

2.8.1. Overview

If you want to get early access to some upcoming features and experiment, you can set the environment variable MINISHIFT_ENABLE_EXPERIMENTAL, which makes additional feature flags available:


Experimental features are not officially supported, and might break or result in unexpected behavior. To share your feedback on these features, you are welcome to contact the Minishift community.

2.8.2. Enabling Experimental oc cluster up Flags

By default, CDK does not expose all oc cluster up flags in the CDK CLI.

You can set the MINISHIFT_ENABLE_EXPERIMENTAL environment variable to enable the following options for the minishift start command:

Installs OpenShift logging aggregation.
Installs OpenShift metrics components.
Enables provisioning the OpenShift service catalog.
Enables passing flags that are not exposed in the CDK CLI directly to oc cluster up.

2.8.3. Set Fixed IP Address


This not currently supported on KVM as the driver plug-in relies on the DHCP offer to determine the IP address.

Most hypervisors do not support extending the lease time when the IP is assigned using DHCP. This might lead to a new IP being assigned to the VM after a restart as it will conflict with the security certificates generated for the old IP. This will make CDK completely unusable until a new instance is set up by running minishift delete followed by minishift start.

To prevent this, CDK includes the functionality to set a static IP address to the VM. This will prevent the IP address from changing between restarts. However, it will not work on all of the driver plug-ins at the moment due to the way the IP address is resolved.

The following command will set the IP address that was assigned as fixed:

$ minishift ip --set-static

If you prefer to use dynamic assignment, you can use:

$ minishift ip --set-dhcp

2.9. Converting an Existing Docker Compose Project

2.9.1. Installing Kompose

Kompose is a tool for converting Docker Compose to container orchestrators such as OpenShift and Kubernetes.

On Windows and macOS, Kompose can be installed through the DevSuite GUI installer. See the Red Hat Development Suite Installation Guide for more information.

On Red Hat Enterprise Linux, Kompose can be installed from the command line by enabling the Red Hat Developer Tools and Red Hat Software Collections repositories:

$ subscription-manager repos --enable rhel-7-server-devtools-rpms
$ subscription-manager repos --enable rhel-server-rhscl-7-rpms
$ yum install kompose -y

2.9.2. Using Kompose

To convert your Docker Compose project using Kompose, follow these steps:

  1. Start Minishift so you have an OpenShift cluster to communicate with.

    $ minishift start
    Starting local OpenShift cluster using 'kvm' hypervisor...
    -- Checking OpenShift client ... OK
    -- Checking Docker client ... OK
    -- Checking Docker version ... OK
    -- Checking for existing OpenShift container ... OK
  2. Download an example Docker Compose file, or use your own.

  3. Convert your Docker Compose file to OpenShift. Run kompose convert --provider=openshift in the same directory as your docker-compose.yaml file.

    $ kompose convert --provider=openshift
    INFO OpenShift file "frontend-service.yaml" created
    INFO OpenShift file "redis-master-service.yaml" created
    INFO OpenShift file "redis-slave-service.yaml" created
    INFO OpenShift file "frontend-deploymentconfig.yaml" created
    INFO OpenShift file "frontend-imagestream.yaml" created
    INFO OpenShift file "redis-master-deploymentconfig.yaml" created
    INFO OpenShift file "redis-master-imagestream.yaml" created
    INFO OpenShift file "redis-slave-deploymentconfig.yaml" created
    INFO OpenShift file "redis-slave-imagestream.yaml" created

    Alternatively, you can convert and deploy directly to OpenShift with kompose up --provider=openshift.

    $ kompose up --provider=openshift
    INFO We are going to create OpenShift DeploymentConfigs, Services and PersistentVolumeClaims for your Dockerized application.
    If you need different kind of resources, use the 'kompose convert' and 'oc create -f' commands instead.
    INFO Deploying application in "myproject" namespace
    INFO Successfully created Service: frontend
    INFO Successfully created Service: redis-master
    INFO Successfully created Service: redis-slave
    INFO Successfully created DeploymentConfig: frontend
    INFO Successfully created ImageStream: frontend
    INFO Successfully created DeploymentConfig: redis-master
    INFO Successfully created ImageStream: redis-master
    INFO Successfully created DeploymentConfig: redis-slave
    INFO Successfully created ImageStream: redis-slave
    Your application has been deployed to OpenShift. You can run 'oc get dc,svc,is,pvc' for details.
  4. Access the newly deployed application with Minishift.

    After deployment, you must create an OpenShift route in order to access the service.

    Create a route for the frontend service using oc.

    $ oc expose service/frontend
    route "frontend" exposed

    Access the frontend service with minishift.

    $ minishift openshift service frontend --namespace=myproject
    Opening the service myproject/frontend in the default browser...

    You can also access the GUI of OpenShift for an overview of the deployed containers.

    $ minishift console
    Opening the OpenShift Web console in the default browser...