Highlights

  • Highlights for the week ending October 15, 2021

    General

    As part of our continuous effort to align our terminology with the Kubernetes project where appropriate, we phase out the term “key pair” and will instead use “client certificate” in the future. Expect this to land in the web UI and documentation soon. Commands in gsctl will not be affected by the change however.

    User interfaces

    We have added changes to kubectl gs:

    • The template cluster and template nodepool commands for release v20.0.0 now use Management API for defaulting, making it the same across providers including Cluster API (for AWS, AWS using EKS, and for Azure).
  • Highlights for the week ending October 8, 2021

    Apps

    nginx-ingress-controller-app v2.3.0 updates the controller image to upstream v1.0.3 to fix problems some users had with a special LUA module.

    Documentation

    Terminology update: what has been called “master nodes” by the Kubernetes project in the past is now called the Control plane or the control plane nodes. We are aligning with these changes in our documentation as well as the user interfaces. Note that we are making an exception for gsctl and the Rest API, which we plan to phase out in the future.

    User interfaces

    We have several important changes in kubectl gs:

    • The template cluster and template nodepool commands will automatically use the namespace of the organization owning the cluster.
    • We added the command get releases to fetch information on cluster releases.
    • The template nodepool command now ensures the giantswarm.io/machine-pool labels for each worker node in a node pool, both on AWS and Azure.

    Users of gsctl will notice deprecation warnings in a growing number of commands with new releases. These will inform you on which kubectl gs command to check out as a replacement. If you are using gsctl in automation, these warnings are printed to STDERR and as we hope should not affect your processing of any STDOUT output.

  • Highlights for the week ending September 17, 2021

    Apps

    Flux App v0.4.0 updates the flux toolkit version to upstream v0.17.1.

    Dashboards

    We welcome two new Grafana dashboards:

    • Managed apps gives you an overview of the apps installed in your workload clusters, plus a extra information on available app updates.

    • AWS cluster status shows some key metrics on your workload clusters. Available only in AWS installations.

    User interfaces

    With kubectl-gs v1.39.1, in the template cluster and template nodepool command, we replaced the flag --owner by --organization. And added the new command template organization for creating an Organization CR.

    In the web interface we improved the performance when browsing apps significantly.

    Documentation

    We updated all our documentation on ingress controllers: the basic guide on installing an ingress controller and all our advanced ingress docs.

  • Highlights for the week ending September 10, 2021

    Apps

    NGINX IC v2.2.0 upgrades to upstream v1.0.0. This is a breaking change! From this version on, only clusters with kubernetes >= 1.19 are supported. The support for Ingress Object in networking.k8s.io/v1beta is being dropped and manifests should now use networking.k8s.io/v1. Please make sure to read the upgrading notes. (#218).

    cert-manager v2.10.0 Upgrade to upstream v1.5.3. This is the first version compatible with Kubernetes 1.22.

    User interfaces

    There is a new Installed Apps view in the Apps tab on the Cluster detail page.

    kubectl-gs v1.39.0 offers a new template organization command for templating Organization CRs.

  • Highlights for the week ending August 27, 2021

    Apps

    Kong v2.3.0 upgrades to upstream chart v2.3.0, which includes Kong v2.5.0 and Kong Ingress Controller v1.3.0. Warning: Breaking change for Kubernetes clusters older than v1.16.x. Please check “Breaking Changes” paragraph in the upstream changelog.

    NGINX IC v2.1.0 upgrades to controller container image v0.49.0.

    User interfaces

    For single sign-on users, we removed the “Helm stable” catalog from the Apps section in the web interface. To make up for it, we drastically improved the loading performance, as all catalog data is now fetched from the Management API.

    kubectl-gs as of v1.36.0 offers an additional flag --callback-server to the login command. If you had trouble with a local firewall or similar before, please consider using this new flag to specify the inbound port to use. This also makes usage in a Docker container easier. Example:

    docker run -p 8080:8080 -ti --rm \
      -v $HOME/.kube:/root/.kube \
      quay.io/giantswarm/kubectl-gs:1.36.0 login \
      --callback-port 8080 \
      https://g8s.your-installation-domain/
    
  • Highlights for the week ending August 20, 2021

    Apps

    Cert Manager v2.9.0 upgrades to upstream v1.4.2, deprecating v1alpha2, v1alpha3, and v1beta1 versions of cert-manager.io and acme.cert-manager.io :warning: If you are upgrading the app on a cluster with Cert Manager upstream < v1.0.0 (Giant Swarm v2.2.5), please make sure Cert Manager CRDs do not reference deprecated APIs. Read upstream release notes.

    External DNS v2.5.0 upgrades to upstream v0.9.0, bringing small improvements and bug fixes.

    User interfaces

    We are continuously adding more guides on using the Management API to the web interface, so you learn to do what you do interactively via a CLI.

    Documentation

    Our new article on observability apps gives you an overview about what we have in store regarding monitoring, logging, graphing and even service mesh.

  • Highlights for the week ending August 6, 2021

    General

    The unique, immutable cluster identifier formerly called the cluster ID is now called the cluster name, and what was called name is now called description. The goal here is consistency between our user interfaces and the Management API. Likewise for node pools.

    Management API

    On AWS, for all cluster-related custom resources we now use version v1alpha3 as the one stored in etcd and the default version served by the Kubernetes API. If you want to obtain the v1alpha2 version, with kubectl you can specify the version like this:

    kubectl get clusters.v1alpha2.cluster.x-k8s.io x1234 -o yaml
    

    Apps

    Linkerd2 v0.6.0, our managed service mesh offering, is now available as an app in the Giant Swarm Catalog. To try it, please read the documentation.

    Prometheus Operator v0.10.0 is upgraded to upstream Kube Prometheus Stack v17.0.3 and now contains prometheus-operator v0.49.0, prometheus v2.28.1 and alertmanager v0.22.2.

    We’ve also improved the monitoring of the managed apps Prometheus Operator, Loki, and Cert Manager for better quality.

    User interfaces

    Deleting a cluster via the web interface (happa v1.15.0) now requires additional confirmation steps, to make sure you are deleting the cluster you want to delete.

    Also in the web interface we start explaining how to achieve via the Management API what you’re doing in the UI. This is only available when using single sign-on (SSO).

    In kubectl gs v1.34.0, the template app command provides options to specify annotations and labels to be applied to an app’s target namespace. Check out the docs for the new flags --namespace-annotations and --namespace-labels.

  • Highlights for the week ending July 23, 2021

    Documentation

    Our CRD docs now feature deprecation info. On the list, there is a DEPRECATED tag. On the details pages, there is a box with details. In cases where the CRD is directly replaced by another one, this info is given in the box.

    Dashboards

    The KVM resource usage dashboard has been added to Grafana.

  • Highlights for the week ending July 16, 2021

    Managed apps

    NGINX IC v2.0.0 is upgraded to controller container image v0.48.1. It contains performance improvements related to the admission webhook. Potentially Breaking change: For security, it disables forwarding traffic to ExternalNameServices by default. To enable it please check the details in the changelog.

    Prometheus Operator v0.9.0 uses giantswarm/config to generate managed configuration. This is part of a larger story to automate the configuration management of installations.

    User interfaces

    In kubectl gs v1.32.0, Catalog CRD replaces AppCatalog CRD, enabling organization-wide (rather than ‘global’), namespace-scoped, app catalogs.

    Documentation

    With release v15.0.0 for Azure you can now specify which external IP address to use for egress traffic of workloads, and we just added documentation for this feature.

  • Highlights for the week ending July 9, 2021

    Managed apps

    We released loki v0.3.0 and promtail v0.3.0 with improvements related to observability and stability of both apps

    User interfaces

    The web interface now features a Monitoring link, as we already mentioned in separate announcement.

    kubectl gs template apps now provides a new flag --app-name to set the name of the app instance. The command also produces much cleaner templates by default.

    Documentation

    To provide for a really simple start with using our App Platform, we added a new Getting started with App Platform guide where you can learn how to deploy apps to your clusters in a declarative way.