Chapter 2. Known Issues

Running an application using Kubernetes fails when a Kubernetes pod needs to use an emptyDir volume because a corresponding SELinux context is not automatically set for emptyDir.

Due to a bug in the vagrant-service-manager plugin, the vagrant service-manager env command reports that the Kubernetes service is stopped in the Red Hat CDK box even when it is running. To work around this issue, users can upgrade to the latest version of the vagrant-service-manager plugin from rubygems.org, in which the bug has been fixed. To upgrade, run the following command: vagrant plugin update vagrant-service-manager or install the newest version by running the following command: vagrant plugin install vagrant-service-manager.

Due to a bug in Atomic CLI, the command to run AtomicApp fails when issued with sudo as follows: sudo atomic run <atomicapp-docker-image>. To work around the problem, use sudo with the -E option: sudo -E atomic run <atomicapp-docker-image>.

Virtualbox is a supported virtualization provider for Red Hat CDK, but the Virtualbox Guest Additions software, which offers various drivers and applications and is useful for sharing folders between the Vagrant box and the host system, is not pre-installed in the Red Hat CDK box because its licence is not compatible with Red Hat CDK. Users are recommended to use SSHFS provided by the vagrant-sshfs plugin to enable Vagrant synchronized folders.

The Red Hat CDK Vagrant box for the Virtualbox virtualization provider does not come with the Virtualbox Guest Additions software pre-installed because of a licensing-incompatibity issue. Users who attempt to work around the absence of Virtualbox Guest Additions by installing the software using the vagrant-vbguest plugin are advised that it is not possible to use both the vagrant-vbguest and vagrant-registartion plugins together. If you wish to install either one of the plugins, make sure that the other one is not installed.

When the vagrant ssh command is used on Mac OS X to log into the Red Hat CDK Vagrant box, the following warning message is displayed:

bash: warning: setlocale: LC_CTYPE: cannot change locale (UTF-8): No such file or directory"

The message can be safely disregarded.

Internal network proxies, which are used for connecting to the Internet in some networking configurations, are not supported by Red Hat CDK.

While it is possible to start the OpenShift service in Red Hat CDK when the host system is not connected to the Internet, Internet connection is required for running applications using OpenShift because OpenShift needs to download required Docker-formatted container images for applications to function.

DNS addressing provided by the external xip.io service is unavailable without a working Internet connection.

Starting or restarting OpenShift takes more time than expected. The reason lies in the way the Red Hat CDK virtual machine is configured. The same problem occurs when running the vagrant service-manager restart openshift command or during initialization using the vagrant up command. This command may fail with a timeout error, even though the environment is running. It is only the OpenShift service that starts slowly.

The Vagrantfile for the rhel-k8s-singlenode-setup configuration does not support credentials stored in the system Vagrantfile. While storing user credentials in the system Vagrantfile works for registering an environment initialized using the rhel-ose configuration, for the rhel-k8s-singlenode-setup Vagranfile, the registration plugin asks for credentials interactively. This problem is caused by a bug in the Vagrantfile.

To work around the problem, delete the config.registration lines from the Vagrantfile, or copy this part from the rhel-ose Vagrantfile.