Release Notes for Dispatch 1.3.0
v1.3.0 - Released October 27, 2020
Document Modified Date: Released October 27, 2020
This document describes the new features, caveats, and resolved issues of D2iQ Dispatch.
- Release notes are sometimes updated with new information about restrictions and caveats. Please see https://docs.d2iq.com/dkp/dispatch/ for the latest version of the D2iQ Dispatch release-notes.
Installation and Upgrades
To install Dispatch, follow one of the two options below:
konvoy initand then update the
Upgrade: Update the
- Update the following example snippet from
cluster.yamlto upgrade Dispatch 1.2.x to 1.3.0.
- configRepository: https://github.com/mesosphere/kubeaddons-dispatch configVersion: stable-1.16-1.2.0 addonsList: - name: dispatch enabled: false
It should read:
- configRepository: https://github.com/mesosphere/kubeaddons-dispatch configVersion: stable-1.18-1.3.0 addonsList: - name: dispatch enabled: true
- Apply the configuration to install/upgrade Dispatch to the Konvoy cluster.
- Verify that the installation/upgrade is successful.
helm test dispatch-kubeaddons
Improvements since v1.2.0
- Upgraded Tekton to v0.14.2 and Tekton dashboard to v0.8.0.
- Upgraded ArgoCD to v1.6.2.
- Added support for Helm v3
- Updated UI to support Build view, including:
- Viewing logs, artifacts, and Dispatchfile.
- Support for stopping, and rerunning Pipelines.
- Support for incremental build numbers of Pipelineruns.
- Default to Dispatch Dashboard URL for PipelineRun Details link. This is configurable by setting
- Support for label matching for pull request actions in Dispatchfile.
timeoutfield to Repository objects. Allows overriding of the default PipelineRun timeout duration for a Repository.
- Added a new
--timeoutflag to the dispatch CI repository create command that takes a Go duration string (e.g.,
2h30m45s). This sets the PipelineRun timeout for the Repository being created.
- Added a new
timeoutfield to the task type in the Dispatchfile. This overrides the PipelineRun timeout for the task. If the PipelineRun times out before the task is completed, the task will continue executing until it reaches its own timeout, but the PipelineRun as a whole will be marked as failed due to timeout.
- Added helm chart option
tekton.configs.defaultPodTemplateto support node pool selections and other pod settings for all Dispatch tasks. See documentation of Tekton pod templates Tekton pod templates for more information.
- Automatically detect the Dispatch installation namespace by checking the labels on the namespace specified by
- Print warning when the CLI and server have different versions.
dispatch ci repository createin order to select which of the
--service-account's SCM secrets to use.
- Allow multiple dispatch login commands to be executed for the same service account.
- Trigger pipelines with GitHub release events.
- Support tolerations via configuring global and per-repository pod template.
- Support tolerations on Dispatch components.
- Support automatic cancellation of pipelineruns superseded by new pipelineruns.
- The helm chart option
eventSink.defaultStepResourcesis renamed to
--dispatch-namespaceCLI option is now deprecated, as its value is automatically determined.
--scm-secretCLI option is now deprecated. Use
--service-accountinstead, and the SCM secret will be automatically determined.
--dispatchfile-repository-scm-secretCLI option is deprecated. The SCM secret for accessing the Dispatchfile is determined from the
--secretoption has been removed from the dispatch login commands, use
- The dispatch init CLI command now uses Helm v3.
- Allow login to github after login to docker. (COPS-6292)
- Fixed a bug where the garbage collector only reclaimed resources in the last watched namespace.
- Fixed garbage collector to only skip old objects with dispatch catalog labels if they are Pipeline or Task objects.
- Fixed a bug where created artifacts were not collected after task failure.
- Fixed Dispatchfile YAML parser to handle quoted “on” keyword correctly.
- Fixed the pipeline generation to handle large webhook event payloads.
- Fixed log pruning job’s permissions in multi-instance mode.
- Added a “Failed” cron status when CrobJob names exceed 52 characters.
- Fixed “Not Supported” error for GitHub release trigger.
The new pipeline auto-cancellation feature automatically cancels pipelines for old commits when new commits are added to your pull request. This can help relieve load on your Kubernetes cluster from unneeded builds. This feature is enabled by default for repositories but each repository can opt out, if needed.
To opt-out, set disableAutoCancel: true on your Repository object, for example:
kind: Repository metadata: name: dispatch spec: repository: mesosphere/dispatch secret: github serviceAccount: builder disableAutoCancel: true
PipelineRun Details link(s) on SCM page now default to the Dispatch UI dashboard. This can be overridden to default to the old behavior of linking to the Tekton dashboard by setting