The Environment Configuration File#

Warning

This page is out of date - the environment.sh varies depending on the type of project:

  • services using compose

  • services using helm

  • deployment

This page needs to be updated to reflect the different types of environment.sh files.

environment.sh is a configuration file that is provided in each domain (beamline or accelerator) repository. It is used to set up the environment such that the epics-containers-cli will interact with the correct cluster (or local server if you are not using Kubernetes).

An important part of creating a new domain repository is to edit the environment.sh so that it suits the domain you are targetting.

There are 3 sections to the file as follows:

Environment Variables Setup#

The first section defines a number of environment variables. These should be adjusted to suit your domain. The variables are as follows:

Required Variables#

  • EC_REGISTRY_MAPPING: defines a mapping between git source repositories and container repositories. This is used to determine where a Generic IOC container image is pushed to (or pulled from). If you are using GitHub then you could use the following value:

    • EC_REGISTRY_MAPPING='github.com=ghcr.io'

    Which means containers whose source repository is at:

    • GitHub/organisation/my_generic_ioc

    will be pushed to:

    • ghcr.io/organisation/my_generic_ioc:VERSION.

    You can have multiple mappings if needed by separating them with a space.

  • EC_K8S_NAMESPACE: defines the namespace in a Kubernetes Cluster that your IOC Instances will be deployed to. When you come to set up a cluster you will need to create a namespace for your domain. This is the name you should use here. If you are not using Kubernetes then you can leave this as EC_K8S_NAMESPACE=local and this will deploy IOC Instances to the local server’s docker or podman instance.

  • EC_SERVICES_REPO: this is a link back to the repository that defines this domain. For example the bl38p reference beamline repository uses EC_SERVICES_REPO=git@github.com:epics-containers/bl38p.git. This variable is used by ec to fetch versioned IOC Instance from the repo and deploy time.

Optional Variables#

  • EC_LOG_URL: if you have a centralized logging service with a web UI then you can set this variable to the URL of the web UI. This will then be displayed when the command ec log-history <ioc-name> is run. The IOC name is added to the URL using {ioc-name} as a placeholder e.g.

    • EC_LOG_URL='https://graylog2.diamond.ac.uk/search?rangetype=relative&fields =message%2Csource&width=1489&highlightMessage=&relative=172800&q=pod_name%3A {ioc_name}*'

  • EC_CONTAINER_CLI: this sets the name of the container CLI to use. supported options are podman or docker. If not set then ec will try to determine which one to use. You would only need this variable if you have both podman and docker installed and you want to use one over the other, or if you want to use a different container CLI such as singularity. IMPORTANT: the application you reference must have docker compatible CLI (at least for common functions).

  • EC_DEBUG: causes the ec command to output debug information for all commands. For more targeted debugging you can use ec -d ....

Installation of ec#

The second section of the environment.sh file is used to install the ec command by pip installing the epics-containers-cli package. The blxxi-template project comes with a suggested way of doing this, but it would probably be best to have ec installed globally on your workstation and then omit this section from your environment.sh files.

Perhaps the simplest way to achieve this is to install ec into your user space using the following command:

pip install --user ec-cli

Then add the following to your $HOME/.bashrc (or $HOME/.zshrc for zsh users):

PATH=$PATH:$HOME/.local/bin

Connecting to a Namespace on your Kubernetes Cluster#

The third section of the environment.sh sets up how the kubectl command will connect to a namespace on your Kubernetes cluster. This usually involves setting the KUBECONFIG environment variable to point to a file that contains the cluster configuration.

When we set up a cluster in the tutorials we will create a namespace for your and discuss how to update environment.sh to connect to it.

If you are connecting to your own facility’s cluster then you will need to ask your admins for the correct configuration.

If you are not using Kubernetes then you can leave this section out.

Note

DLS users: the module system is used to connect us to each beamline/accelerator cluster. The example environment.sh file in bl38p shows how to do this see epics-containers/bl38p

One other thing that is useful is to set up command line completion for the Kubernetes tools kubectl and helm. See the end of the bl38p environment.sh file for an example of how to do this at epics-containers/bl38p.

On this page
Edit on GitHub
Show Source