How to review installation of tools targeting the Lab

The Reviewer is (in most cases) a Developer/Component Owner who wants to test if the software they specified to be built for the Lab is installed and functioning properly both stand-alone and in conjunction with other software in case this is required (e.g. co-simulation).

The steps that the Reviewer needs to follow to assess the installation that is being managed with the Spack Package manager resemble those that the users follow. The difference is that the Reviewers can connect to two different Collaboratory Lab instances to validate the installations for different purposes.

Lab Instances

Instance	Description
The Lab Integration instance (lab-int)	This is the integration environment of the Collaboratory Lab. Reviewers can connect there to test software prior to its deployment to the Lab Production instance.
The Lab Production instance (lab-prod)	This is the production environment of the Collaboratory Lab, i.e., the one that Users work with. Reviewers can connect to the production environment to validate a new release of the EBRAINS tools after its deployment to the production Lab instance but before it is activated by default for EBRAINS users. This ensures the build process in production has no functional errors as a last form of testing. Activation of the new release to be tested before it is available to all the Users will require a few simple manual steps documented below.

Instance

Description

The Lab Integration instance (lab-int)

This is the integration environment of the Collaboratory Lab. Reviewers can connect there to test software prior to its deployment to the Lab Production instance.

The Lab Production instance (lab-prod)

This is the production environment of the Collaboratory Lab, i.e., the one that Users work with. Reviewers can connect to the production environment to validate a new release of the EBRAINS tools after its deployment to the production Lab instance but before it is activated by default for EBRAINS users. This ensures the build process in production has no functional errors as a last form of testing. Activation of the new release to be tested before it is available to all the Users will require a few simple manual steps documented below.

Now we will document how a Reviewer can connect to the aforementioned Lab instances in order to conduct the necessary software validation

lab-int

Login to the Lab integration instance

Firstly, reviewers need to login to the Lab integration instance. Only for the first time accessing the lab-int, to successfully login, reviewers will need to:

create an int account on the iam-int
login with the new int account to drive-int once, to set the appropriate permissions.

Select Image

Select the Official EBRAINS Docker image 22.07 for Collaboratory.Lab (recommended) docker image from the list of imagesand start theserver.

Select Notebook

At the Launcher’s screen, select to start a new Notebook with the kernel that needs to be tested

When a new notebook is started, any software that needs to be tested can be seamlessly loaded through standard python “import” commands:

lab-prod

Start a Lab Session

From a collab, start a Lab session selecting a site to run this Lab session.

Select the Lab Image

Select the Official EBRAINS Docker image 22.07 for Collaboratory.Lab (recommended) docker image from the list of images and start theserver.

Test Lab Kernel

The JupyterLab kernel that needs to be tested is not yet publicly available, so it needs to be loaded manually.

At the Launcher’s screen select to start a new Terminal. In the terminal, issue the followingcommand:

Note: All the necessary variable substitutions will be communicated to the Reviewers each time a new release needs to be tested.)

Then in JupyterLab, click File > New Launcher and the kernel that was loaded manually should be available for selection.

The Launcher’s Screen

At the Launcher’s screen, select to start a new Notebook with that kernel

Import Package

When a new notebook is started, any software that needs to be tested can be seamlessly loaded through standard python “import” commands: