Announcing SIGHUP Distribution v1.36.0

Announcing SIGHUP Distribution v1.36.0
Announcing SIGHUP Distribution v1.36.0

A new Kubernetes upstream release means new APIs to support, new components to test, and, inevitably, a few things to rethink along the way.

We are pleased to announce the release of SIGHUP Distribution 1.36.0 which brings Kubernetes 1.36 support, introduces a brand-new user interface and marks a milestone for the Immutable provider.

We made significant changes under the surface too.

Since v1.35.1, we've merged 20 pull requests in the Distribution repository and more than 40 in the CLI. Our tool Gatekeeper Policy Manager had a major upgrade to v2 and this new version is officially feature-complete.

Kubernetes and module updates

As a CNCF certified distribution, we maintain multiple Kubernetes versions. As usual, our development cycle covers three minor Kubernetes releases, and to do that, in addition to SIGHUP Distribution 1.36.0, we have also released two other supported versions: 1.35.2 and 1.34.3.

OnPremises and Immutable clusters now run Kubernetes v1.36.5. We've also validated the EKS installer against 1.36 and updated every core module for all the new releases.

A new Kubernetes UI

Checking the state of a cluster sometimes means jumping between different tools. Prometheus for alerts, Velero for backups, something else again for logs and so on. Each of these components does its job. The problem is that you're the one stitching them together.

v1.36.0 introduces the SIGHUP Distribution UI to do some of that for you. We built it on Headlamp, every feature you'll see below is a custom open-source Headlamp plugin.

SD UI, the user interface of SIGHUP Distribution, offers an extensible dashboard packed with useful features:

Cluster overview

The SD UI home page: cluster details, module status and summaries

Logs

Logs from Loki, filtered by namespace, workload and pod

Alerts

Firing Prometheus alerts

Backups

Velero backups and schedules

Want to try it now? OnPremises and Immutable kinds can activate Headlamp in their furyctl.yaml. The new core Utilities module documentation covers all the use cases.

We plan to enhance the dashboard with more capabilities and show more cluster insights. Stay tuned!

Immutable moves to beta

We introduced the new Immutable provider in alpha with SD v1.35, which was able to provision Kubernetes on bare-metal servers or VMs with Flatcar Container Linux. It's now moving into beta, with a structured upgrade path for Kubernetes, Flatcar on the nodes and the load balancers.

Cluster nodes are upgraded one at a time. On each node the installer stages and verifies the exact Flatcar version pinned by SD prior to draining and rebooting it. If the update is unavailable or fails verification, the upgrade stops before draining the node and allows it to continue running its workloads. Once the node is uncordoned the upgrade will wait for its pods to become Ready before proceeding. The Immutable installer release notes describe all the progress made.

Gatekeeper Policy Manager v2

The policy module enhances GPM from version v1.1.1 to v2.1.0 by releasing version 2.0.0. The backend is completely rewritten in Go with a smaller container image and reduced dependencies. All pages you are using will continue to work as they did before.

Version 2.0.0 includes a summary in the homepage, single violation links, alerts about incomplete policy enforcement, and Mutation and Gatekeeper events views. Version 2.1.0 offers a new view called Resources view, which works from your objects rather than the policies and with a new RBAC-aligned mode GPM can show the user only the resources that they have access to. The release notes list all the changes.

The new GPM v2.1.0 overview

If you have customized your deployment, for example by turning on OIDC authentication or showing multiple clusters, please follow the v2.0.0 upgrade process.

furyctl UX improvements

There used to be a case where running furyctl apply could have required a second try. When the default StorageClass was not available at the time of installing an application that required storage, the apply could fail despite the StorageClass becoming available at a later stage.

Not ideal.

Starting from v0.36.0, furyctl automatically reruns the distribution phase when needed. This allows the apply operation to complete once the StorageClass becomes available.

We've also changed how furyctl displays progress during an apply.

furyctl now shows only the last lines produced by each step and clears them before going to the next one. That way, you will see only the information relevant to what is being done at the moment.

furyctl apply rerun and output enhancement

Failure messages have also gotten some improvements. With furyctl v0.36.1, failure messages show the actual reason for a failure, including unreachable hosts info. Then you will see the command that failed and a small snippet of its output. A complete version of the output remains logged to a file.

Beyond these changes, there are two other new subcommands in furyctl v0.36.0.

The first is furyctl create secrets, which creates secrets from the CLI directly, including SSO-related secrets, and can even add references to these created files in your furyctl.yaml.

The second one is furyctl renew kubeconfigs. It makes life easier by letting you renew admin and user kubeconfigs, or even generate one for a newly added user, without having to rerun the Kubernetes phase.

Wrapping it up

The complete SIGHUP Distribution v1.36.0 changelog is available here. This version has breaking changes and a known Cilium issue; read the release notes before upgrading to see whether your cluster is affected.

We welcome any suggestions from the community and are excited about what is coming next. If you are curious about how the internals work and how it can help your team, reach out!