Flexible workloads list in Backstage

[marians]

In our Backstage UI, we want to provide an overview of workloads throughout clusters, with information that is gathered exclusively on Giant Swarm management clusters. As of now (Jan 2025) this covers workloads deployed via App and HelmRelease resources.

The tabular overview should provide the most important details per workload. User-selectable columns might offer more details. In addition users will be able to open a details panel or navigate to a details page for more information on the individual deployment.

Filter functions should be offered to select workloads of interest. For example:

• Instances of selected applications
• Workloads of a certain type (e. g. App, HelmRelease)
• Workloads deployed to certain types of clusters (e. g. Kubernetes version, prod clusters)
• Workloads deployed to certain cloud providers or regions

Improvements/Iterations

Preview Give feedback

1. Allow for more (50, 100) table items per page
2. Sort table by name (desc) by default
3. Ensure correct cluster type for App CRs deployed via bundles

[marians]

Access to wireframes

[marians]

As we introduce a new main menu item, we need at least one new icon.

• Clusters: import StorageIcon from '@mui/icons-material/Storage';
• Deployments: import CloudUploadIcon from '@mui/icons-material/CloudUpload';
• Installations: import FactoryIcon from '@mui/icons-material/Factory';

[marians]

In case of HelmRelease resources, the SOURCE column currently shows two different types of info.

• HelmRepository (obsolete)
• Bucket (unused)
• GitRepository
• OCIRepository

@piontec will provide some more examples

[marians]

One particular case seems to be untreated currently: the fact that an app is deployed via a bundle is not visible at all in our deployments table or details.

For example, this devportal page lists deployment of athena. In the gazelle/operations cluster, it is deployed via the auth-bundle, so it cannot be upgraded individually.

Let's discuss what we could add to make that more visible.

[gusevda]

We can probably have some relationships graph on deployment details page (that we currently don't have). In the deployments table I think it's a good idea to emphasize the fact that's an app is part of a bundle.

[gusevda]

A new page with overview of all apps deployed throughout clusters was added - https://devportal.giantswarm.io/deployments
There is currently a known issue when CLUSTER TYPE column can display incorrect information. The reason is in the logic where we identify if an App CR targets a management cluster or a workload cluster. We are doing it by looking at .spec.kubeConfig.inCluster flag, but recently we found out that it doesn't work the same way for App CRs that are part of an app bundle. We will try to fix this issue in the following releases.

[marians]

One possible improvement: allow for more than 20 rows per page. 50, 100 would be good next choices.

[marians]

I can't really determine the sorting behaviour of the table in the current implemenentation. Is it unsorted?

Ideally by default it would be sorted by NAME, in descending order, so that the arrow indicator points downwards, like here: