fix(manifests): fix full CRDs and update tests to use them. Fixes #8532…

… (#14044)

Signed-off-by: Mason Malone <[email protected]>

docs/executor_swagger.md

@@ -983,7 +983,7 @@ ConfigMap volumes support ownership management and SELinux relabeling.
 | template | string| `string` |  | | Name of template to execute |  |
 | templateRef | [TemplateRef](#template-ref)| `TemplateRef` |  | |  |  |
 | when | string| `string` |  | | When is an expression in which the task should conditionally execute |  |
-| withItems | [][Item](#item)| `[]Item` |  | | WithItems expands a task into multiple parallel tasks from the items in the list |  |
+| withItems | [][Item](#item)| `[]Item` |  | | WithItems expands a task into multiple parallel tasks from the items in the list</br>Note: The structure of WithItems is free-form, so we need</br>"x-kubernetes-preserve-unknown-fields: true" in the validation schema.</br>+kubebuilder:validation:Schemaless</br>+kubebuilder:pruning:PreserveUnknownFields |  |
 | withParam | string| `string` |  | | WithParam expands a task into multiple parallel tasks from the value in the parameter,</br>which is expected to be a JSON list. |  |
 | withSequence | [Sequence](#sequence)| `Sequence` |  | |  |  |
 
@@ -1856,7 +1856,6 @@ ISCSI volumes support ownership management and SELinux relabeling.
 
 
 > +protobuf.options.(gogoproto.goproto_stringer)=false
-+kubebuilder:validation:Type=object
   
 
 
@@ -2511,11 +2510,8 @@ be cluster-scoped, so there is no namespace field.
 ### <span id="parallel-steps"></span> ParallelSteps
 
 
-> +kubebuilder:validation:Type=array
 
 
-
-
 [interface{}](#interface)
 
 ### <span id="parameter"></span> Parameter
@@ -3391,7 +3387,7 @@ cause implementors to also use a fixed point implementation.
 | resources | [ResourceRequirements](#resource-requirements)| `ResourceRequirements` |  | |  |  |
 | restartPolicy | [ContainerRestartPolicy](#container-restart-policy)| `ContainerRestartPolicy` |  | |  |  |
 | securityContext | [SecurityContext](#security-context)| `SecurityContext` |  | |  |  |
-| source | string| `string` |  | | Source contains the source code of the script to execute |  |
+| source | string| `string` |  | | Source contains the source code of the script to execute</br>+optional |  |
 | startupProbe | [Probe](#probe)| `Probe` |  | |  |  |
 | stdin | boolean| `bool` |  | | Whether this container should allocate a buffer for stdin in the container runtime. If this</br>is not set, reads from stdin in the container will always result in EOF.</br>Default is false.</br>+optional |  |
 | stdinOnce | boolean| `bool` |  | | Whether the container runtime should close the stdin channel after it has been opened by</br>a single attach. When stdin is true the stdin stream will remain open across multiple attach</br>sessions. If stdinOnce is set to true, stdin is opened on container start, is empty until the</br>first client attaches to stdin, and then remains open and accepts data until the client disconnects,</br>at which time stdin is closed and remains closed until the container is restarted. If this</br>flag is false, a container processes that reads from stdin will never receive an EOF.</br>Default is false</br>+optional |  |

docs/fields.md

@@ -1743,7 +1743,7 @@ Template is a reusable and composable unit of execution in a workflow
 |`nodeSelector`|`Map< string , string >`|NodeSelector is a selector to schedule this step of the workflow to be run on the selected node(s). Overrides the selector set at the workflow level.|
 |`outputs`|[`Outputs`](#outputs)|Outputs describe the parameters and artifacts that this template produces|
 |`parallelism`|`integer`|Parallelism limits the max total parallel pods that can execute at the same time within the boundaries of this template invocation. If additional steps/dag templates are invoked, the pods created by those templates will not be counted towards this total.|
-|`plugin`|[`Plugin`](#plugin)|Plugin is a plugin template|
+|`plugin`|[`Plugin`](#plugin)|Plugin is a plugin template Note: the structure of a plugin template is free-form, so we need to have "x-kubernetes-preserve-unknown-fields: true" in the validation schema.|
 |`podSpecPatch`|`string`|PodSpecPatch holds strategic merge patch to apply against the pod spec. Allows parameterization of container fields which are not strings (e.g. resource limits).|
 |`priority`|`integer`|Priority to apply to workflow pods.|
 |`priorityClassName`|`string`|PriorityClassName to apply to workflow pods.|
@@ -3263,13 +3263,13 @@ WorkflowStep is a reference to a template to execute in a series of step
 |`arguments`|[`Arguments`](#arguments)|Arguments hold arguments to the template|
 |`continueOn`|[`ContinueOn`](#continueon)|ContinueOn makes argo to proceed with the following step even if this step fails. Errors and Failed states can be specified|
 |`hooks`|[`LifecycleHook`](#lifecyclehook)|Hooks holds the lifecycle hook which is invoked at lifecycle of step, irrespective of the success, failure, or error status of the primary step|
-|`inline`|[`Template`](#template)|Inline is the template. Template must be empty if this is declared (and vice-versa).|
+|`inline`|[`Template`](#template)|Inline is the template. Template must be empty if this is declared (and vice-versa). Note: This struct is defined recursively, since the inline template can potentially contain steps/DAGs that also has an "inline" field. Kubernetes doesn't allow recursive types, so we need "x-kubernetes-preserve-unknown-fields: true" in the validation schema.|
 |`name`|`string`|Name of the step|
 |~~`onExit`~~|~~`string`~~|~~OnExit is a template reference which is invoked at the end of the template, irrespective of the success, failure, or error of the primary template.~~ DEPRECATED: Use Hooks[exit].Template instead.|
 |`template`|`string`|Template is the name of the template to execute as the step|
 |`templateRef`|[`TemplateRef`](#templateref)|TemplateRef is the reference to the template resource to execute as the step.|
 |`when`|`string`|When is an expression in which the step should conditionally execute|
-|`withItems`|`Array<`[`Item`](#item)`>`|WithItems expands a step into multiple parallel steps from the items in the list|
+|`withItems`|`Array<`[`Item`](#item)`>`|WithItems expands a step into multiple parallel steps from the items in the list Note: The structure of WithItems is free-form, so we need "x-kubernetes-preserve-unknown-fields: true" in the validation schema.|
 |`withParam`|`string`|WithParam expands a step into multiple parallel steps from the value in the parameter, which is expected to be a JSON list.|
 |`withSequence`|[`Sequence`](#sequence)|WithSequence expands a step into a numeric sequence|
 
@@ -4020,13 +4020,13 @@ DAGTask represents a node in the graph during DAG execution
 |`dependencies`|`Array< string >`|Dependencies are name of other targets which this depends on|
 |`depends`|`string`|Depends are name of other targets which this depends on|
 |`hooks`|[`LifecycleHook`](#lifecyclehook)|Hooks hold the lifecycle hook which is invoked at lifecycle of task, irrespective of the success, failure, or error status of the primary task|
-|`inline`|[`Template`](#template)|Inline is the template. Template must be empty if this is declared (and vice-versa).|
+|`inline`|[`Template`](#template)|Inline is the template. Template must be empty if this is declared (and vice-versa). Note: As mentioned in the corresponding definition in WorkflowStep, this struct is defined recursively, so we need "x-kubernetes-preserve-unknown-fields: true" in the validation schema.|
 |`name`|`string`|Name is the name of the target|
 |~~`onExit`~~|~~`string`~~|~~OnExit is a template reference which is invoked at the end of the template, irrespective of the success, failure, or error of the primary template.~~ DEPRECATED: Use Hooks[exit].Template instead.|
 |`template`|`string`|Name of template to execute|
 |`templateRef`|[`TemplateRef`](#templateref)|TemplateRef is the reference to the template resource to execute.|
 |`when`|`string`|When is an expression in which the task should conditionally execute|
-|`withItems`|`Array<`[`Item`](#item)`>`|WithItems expands a task into multiple parallel tasks from the items in the list|
+|`withItems`|`Array<`[`Item`](#item)`>`|WithItems expands a task into multiple parallel tasks from the items in the list Note: The structure of WithItems is free-form, so we need "x-kubernetes-preserve-unknown-fields: true" in the validation schema.|
 |`withParam`|`string`|WithParam expands a task into multiple parallel tasks from the value in the parameter, which is expected to be a JSON list.|
 |`withSequence`|[`Sequence`](#sequence)|WithSequence expands a task into a numeric sequence|
 

docs/installation.md

@@ -25,6 +25,16 @@ You can use Kustomize to patch your preferred [configurations](managed-namespace
 
 You can install Argo Workflows using the community maintained [Helm charts](https://github.com/argoproj/argo-helm).
 
+### Full CRDs
+
+The official release manifests come with stripped-down CRDs that omit validation information.
+This is a workaround for [Kubernetes size limitations](https://github.com/kubernetes/kubernetes/issues/82292) when using client-side apply.
+As of version 3.7, the full CRDs can be installed using [server-side apply](https://kubernetes.io/docs/reference/using-api/server-side-apply/) via the following command:
+
+```bash
+kubectl apply --server-side --kustomize https://github.com/argoproj/argo-workflows/manifests/base/crds/full?ref=v3.7.0
+```
+
 ## Installation options
 
 Determine your base installation option.

hack/manifests/crdgen.sh

@@ -8,7 +8,7 @@ add_header() {
   mv tmp "$1"
 }
 
-controller-gen crd:maxDescLen=0 paths=./pkg/apis/... output:dir=manifests/base/crds/full
+controller-gen crd:maxDescLen=0,generateEmbeddedObjectMeta=true paths=./pkg/apis/... output:dir=manifests/base/crds/full
 
 find manifests/base/crds/full -name 'argoproj.io*.yaml' | while read -r file; do
   # remove junk fields