Jan 28, 2025 - hsk8s 3.3.0 for forms v41+
Today we release version 3.3.0 of our hsk8s CLI tool.
Required for v41 and later
Make sure to download
hsk8s3.3.0 before streaming images in preparation for an upgrade to v41 or later.
With forms version 41 and later, hsk8s will now also stream the latest available block images, even if they are not included in the latest version of Hyperscience.
For example, if you are upgrading from version 40.0.3 to 41.0.4, using hsk8s image stream --forms-version 41.0.4 will make version 40.2.6 of TABLE_LOCATOR_4 available, while also introducing TABLE_LOCATOR_5, which is included in version 41.0.4. That way, you can receive patch fixes and improvements for blocks that were released not only in the latest version, but also in the two previous major versions.
Dec 15, 2024 - hsk8s 3.2.1 and Helm Chart 9.0 Release
Today we release the following versions of hsk8s and our Helm Chart:
Version 3.2.1 of our hsk8s CLI tool — Updates the default trainer Postgres image streamed to version 16.6. See hsk8s (Hyperscience Kubernetes CLI).
Version 9.0.0 of our Helm Chart — Updates the default trainer Postgres image used to version 16.6. Please ensure you stream the new Postgres image using hsk8s version >= 3.2.1 before upgrading the Helm Chart.
Oct 9, 2023 - hsk8s 3.1.0 Release (Feature)
Today we release version 3.1.0 of our hsk8s CLI tool.
Added asset management commands. Adds ability to list, download, stream, and import block assets. See hsk8s docs.
Sep 19, 2023 - Helm Chart 8.0 Release
Although this release has no breaking changes, extra steps are needed to enable new functionality.
First, we've introduced a new CRD, the hyperscaler. It handles block and trainer autoscaling. For more information on how to set it up, please check the Scaling article.
Second, we've made minor changes to the hypertrainer CRD. Since Helm doesn't automatically update them, the update needs to be completed manually:
helm repo update
helm template hyperscience/hyperscience --include-crds -s hypertrainer.yaml | kubectl apply -f -Mar 31, 2023 - Helm Chart 7.0 Release (Breaking Change)
Today we release version 7.0.0 of our Helm Chart.
This release introduces 2 breaking changes
serviceAccount blocks are now in place for blocks: , trainer: , app: and operator:
You will need to migrate your existing serviceAccount block to app.serviceAccount. If you experience issues with accessing object storage after upgrading to this version, this may be the reason why.
The second change is required only if you have oidc.enabled: true.
This version removes the following settings from the values.yaml file:
oidc.existingClientSecretNameoidc.clientIdoidc.clientSecret
If you had any of them set, please instead add the HS_OIDC_RP_CLIENT_ID and HS_OIDC_RP_CLIENT_SECRET keys to the platform secret. The name of the secret can be found at secrets.platform in your values file.
Mar 10, 2023 - hsk8s 3.0.0 Release (Breaking Change)
Today we release version 3.0.0 of our hsk8s CLI tool.
Added support for Windows OS.
This version does not support v1alpha1 version for k8s user authentication. The last hsk8s version that supports it is 2.8.0.
Compiled with golang v1.20.
Oct 6, 2022 - Helm Chart 6.0 Release (Breaking Change)
Today we release version 6.0.0 of our Helm Chart.
This version defaults to Operator version 4.0.0, which is compatible only with Platform versions 33.1.32+, 34.0.9+ and 35.0.0+.
This operator version fixes a critical bug with how Hyperscience blocks are started, and enables Python Code blocks to work on Kubernetes deployments.
If your deployment of the Hyperscience Platform is on an older version, make sure to use Operator version 3.4.2.
Sep 21, 2022 - Helm Chart 5.0 Release (Breaking Change)
Today we release version 5.0.0 of our Helm Chart. Until now block variables were specified under the operator map.
Starting with version 5.0.0 all blocks the following breaking changes in the values.yaml file are introduced:
before -> after
operator.sdmRepository: "" -> blocks.repository: ""
operator.pollInterval: 10 -> blocks.pollInterval: 10This will allow us to continue adding new features and make it clear that they are specific to blocks.
In addition to this we have also added support for kubernetes tolerations for all resources. To support this hyperoperator 3.4.0 is also released and used by default in helm chart v5.0.0.
Aug 12, 2022 -- Helm Chart 4.0 Release (Breaking Change)
With the release of Helm Chart version 4.0, it is required to have one more Docker repository, by default called trainer. It will be used to store the trainer application component. Up until now, both applications were using the same image. Starting from application version 35.0.0, the trainer and forms Docker images will not be the same, allowing us to reduce the size of each image.
The trainer image can be streamed to your internal registry with the hsk8s image stream --trainer subcommand, added in hsk8s version 2.1.0.
All forms versions now have a matching trainer image that you can stream, including versions prior to 35.0.0.
The new required trainer repository URL can be configured in your values.yaml file, as follows:
trainer:
repository: "my-private-registry/trainer" # required
tags:
- 34.0.5
June 23, 2022 -- Helm Chart 3.0 Release (Breaking Change)
Today, we released version 3.0.0 of our Helm Chart, which now includes the AWS RDS SSL cert bundle in a ConfigMap. This ConfigMap is mounted to /etc/nginx/certs in all frontend and backend pods by default.
Upgrading to this version of the Helm Chart may cause errors on helm upgrade if you have previously configured the app's ConfigMap mounts (in values.yaml) to include a cert that is managed outside of the helm chart and mounted to /etc/nginx/certs. See the below code block as an example:
app:
dotenv:
configMapMounts:
# If you have configured a configMapMount that
# is mounted to /etc/nginx/certs, you may need to remove this
# block from your values.yaml
- name: "rds-cert"
mountPath: "/etc/nginx/certs"
configMap: "aws-rds-cert"
readOnly: trueIf you wish to exclude the AWS RDS SSL cert bundle from your application's volume mounts, you can do so by adding the following snippet to your values.yaml file:
cloud:
aws:
includeRdsCerts: falseJune 6, 2022 -- Consolidated Block Image Repository
We have released new versions of our tooling for installing Hyperscience in on-premise Kubernetes (k8s) clusters. With these new releases, we have consolidated all workflow (SDM) block images into a single repository! Previously, customers were responsible for managing a large set of Docker Repositories that could potentially change during system upgrades. Now, customers will only need to manage a total of three Docker Repositories: one for the main application (forms), one for the Hyperscience K8s Operator (hyperoperator), and one for all Workflow Blocks (sdm_blocks). Changes have been made to our helm chart and k8s operator to support this new paradigm, and we will no longer be supporting the previous block-image-management scheme.
These changes are reflected in the new 2.0 release of our command-line k8s tool, hsk8s. While most of the functionality has remained the same, in 2.0.0, you can now inspect which blocks are available for each application version using the hsk8s image list-blocks or hsk8s image block-versions subcommands. The hsk8s image stream subcommand will continue to stream Hyperscience Docker images to their appropriate repositories (now assuming that the 3 required repositories are present in your Docker Registry). Furthermore, improvements have been made around the customizability of this command through command line flags, which are documented in the tool itself by typing hsk8s image stream --help. These features should improve your experience when installing and supporting Hyperscience in your own on-premise k8s clusters.
Required Components and Versions
Component | Version |
|---|---|
hsk8s | 2.0+ |
hyperoperator | 2.0+ |
helm chart | 2.4+ |