root
/
kubernetes
mirror of https://github.com/kubernetes/kubernetes.git
1
0
Fork 0
Code Issues Actions Packages Projects Releases Wiki Activity
kubernetes/pkg/apis/resource/validation/validation_resourceslice_te...

544 lines
27 KiB
Go
Raw Normal View History

dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
/*
Copyright 2022 The Kubernetes Authors.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
*/
package validation
import (
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
"fmt"
"strings"
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
"testing"
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
"k8s.io/apimachinery/pkg/api/resource"
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
"k8s.io/apimachinery/pkg/util/validation/field"
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
"k8s.io/kubernetes/pkg/apis/core"
resourceapi "k8s.io/kubernetes/pkg/apis/resource"
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
"k8s.io/utils/ptr"
)
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
func testAttributes() map[resourceapi.QualifiedName]resourceapi.DeviceAttribute {
return map[resourceapi.QualifiedName]resourceapi.DeviceAttribute{
"int": {IntValue: ptr.To(int64(42))},
"string": {StringValue: ptr.To("hello world")},
"version": {VersionValue: ptr.To("1.2.3")},
"bool": {BoolValue: ptr.To(true)},
}
}
DRA API: use DeviceCapacity struct instead of plain Quantity This enables a future extension where capacity of a single device gets consumed by different claims. The semantic without any additional fields is the same as before: a capacity cannot be split up and is only an attribute of a device. Because its semantically the same as before, two-way conversion to v1alpha3 is possible.
2024-09-26 22:56:48 +08:00
func testCapacity() map[resourceapi.QualifiedName]resourceapi.DeviceCapacity {
return map[resourceapi.QualifiedName]resourceapi.DeviceCapacity{
DRA API: rename DeviceCapacity.Quantity to DeviceCapacity.Value Based on review feedback (https://github.com/kubernetes/kubernetes/pull/127511#discussion_r1823521172).
2024-11-01 04:30:28 +08:00
"memory": {Value: resource.MustParse("1Gi")},
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
}
}
func testResourceSlice(name, nodeName, driverName string, numDevices int) *resourceapi.ResourceSlice {
slice := &resourceapi.ResourceSlice{
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
ObjectMeta: metav1.ObjectMeta{
Name: name,
},
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
Spec: resourceapi.ResourceSliceSpec{
DRA: new API for 1.31 This is a complete revamp of the original API. Some of the key differences: - refocused on structured parameters and allocating devices - support for constraints across devices - support for allocating "all" or a fixed amount of similar devices in a single request - no class for ResourceClaims, instead individual device requests are associated with a mandatory DeviceClass For the sake of simplicity, optional basic types (ints, strings) where the null value is the default are represented as values in the API types. This makes Go code simpler because it doesn't have to check for nil (consumers) and values can be set directly (producers). The effect is that in protobuf, these fields always get encoded because `opt` only has an effect for pointers. The roundtrip test data for v1.29.0 and v1.30.0 changes because of the new "request" field. This is considered acceptable because the entire `claims` field in the pod spec is still alpha. The implementation is complete enough to bring up the apiserver. Adapting other components follows.
2024-06-18 23:47:29 +08:00
NodeName: nodeName,
Driver: driverName,
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
Pool: resourceapi.ResourcePool{
DRA: new API for 1.31 This is a complete revamp of the original API. Some of the key differences: - refocused on structured parameters and allocating devices - support for constraints across devices - support for allocating "all" or a fixed amount of similar devices in a single request - no class for ResourceClaims, instead individual device requests are associated with a mandatory DeviceClass For the sake of simplicity, optional basic types (ints, strings) where the null value is the default are represented as values in the API types. This makes Go code simpler because it doesn't have to check for nil (consumers) and values can be set directly (producers). The effect is that in protobuf, these fields always get encoded because `opt` only has an effect for pointers. The roundtrip test data for v1.29.0 and v1.30.0 changes because of the new "request" field. This is considered acceptable because the entire `claims` field in the pod spec is still alpha. The implementation is complete enough to bring up the apiserver. Adapting other components follows.
2024-06-18 23:47:29 +08:00
Name: nodeName,
ResourceSliceCount: 1,
},
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
},
}
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
for i := 0; i < numDevices; i++ {
device := resourceapi.Device{
Name: fmt.Sprintf("device-%d", i),
Basic: &resourceapi.BasicDevice{
Attributes: testAttributes(),
Capacity: testCapacity(),
},
}
slice.Spec.Devices = append(slice.Spec.Devices, device)
}
return slice
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
}
dra api: rename NodeResourceSlice -> ResourceSlice While currently those objects only get published by the kubelet for node-local resources, this could change once we also support network-attached resources. Dropping the "Node" prefix enables such a future extension. The NodeName in ResourceSlice and StructuredResourceHandle then becomes optional. The kubelet still needs to provide one and it must match its own node name, otherwise it doesn't have permission to access ResourceSlice objects.
2024-03-07 17:14:11 +08:00
func TestValidateResourceSlice(t *testing.T) {
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
goodName := "foo"
badName := "!@#$%^"
driverName := "test.example.com"
now := metav1.Now()
badValue := "spaces not allowed"
scenarios := map[string]struct {
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
slice *resourceapi.ResourceSlice
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
wantFailures field.ErrorList
}{
"good": {
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
slice: testResourceSlice(goodName, goodName, driverName, resourceapi.ResourceSliceMaxDevices),
},
"too-large": {
wantFailures: field.ErrorList{field.TooMany(field.NewPath("spec", "devices"), resourceapi.ResourceSliceMaxDevices+1, resourceapi.ResourceSliceMaxDevices)},
slice: testResourceSlice(goodName, goodName, goodName, resourceapi.ResourceSliceMaxDevices+1),
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
},
"missing-name": {
wantFailures: field.ErrorList{field.Required(field.NewPath("metadata", "name"), "name or generateName is required")},
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
slice: testResourceSlice("", goodName, driverName, 1),
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
},
"bad-name": {
wantFailures: field.ErrorList{field.Invalid(field.NewPath("metadata", "name"), badName, "a lowercase RFC 1123 subdomain must consist of lower case alphanumeric characters, '-' or '.', and must start and end with an alphanumeric character (e.g. 'example.com', regex used for validation is '[a-z0-9]([-a-z0-9]*[a-z0-9])?(\\.[a-z0-9]([-a-z0-9]*[a-z0-9])?)*')")},
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
slice: testResourceSlice(badName, goodName, driverName, 1),
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
},
"generate-name": {
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
slice: func() *resourceapi.ResourceSlice {
slice := testResourceSlice(goodName, goodName, driverName, 1)
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
slice.GenerateName = "prefix-"
return slice
}(),
},
"uid": {
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
slice: func() *resourceapi.ResourceSlice {
slice := testResourceSlice(goodName, goodName, driverName, 1)
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
slice.UID = "ac051fac-2ead-46d9-b8b4-4e0fbeb7455d"
return slice
}(),
},
"resource-version": {
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
slice: func() *resourceapi.ResourceSlice {
slice := testResourceSlice(goodName, goodName, driverName, 1)
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
slice.ResourceVersion = "1"
return slice
}(),
},
"generation": {
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
slice: func() *resourceapi.ResourceSlice {
slice := testResourceSlice(goodName, goodName, driverName, 1)
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
slice.Generation = 100
return slice
}(),
},
"creation-timestamp": {
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
slice: func() *resourceapi.ResourceSlice {
slice := testResourceSlice(goodName, goodName, driverName, 1)
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
slice.CreationTimestamp = now
return slice
}(),
},
"deletion-grace-period-seconds": {
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
slice: func() *resourceapi.ResourceSlice {
slice := testResourceSlice(goodName, goodName, driverName, 1)
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
slice.DeletionGracePeriodSeconds = ptr.To[int64](10)
return slice
}(),
},
"owner-references": {
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
slice: func() *resourceapi.ResourceSlice {
slice := testResourceSlice(goodName, goodName, driverName, 1)
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
slice.OwnerReferences = []metav1.OwnerReference{
{
APIVersion: "v1",
Kind: "pod",
Name: "foo",
UID: "ac051fac-2ead-46d9-b8b4-4e0fbeb7455d",
},
}
return slice
}(),
},
"finalizers": {
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
slice: func() *resourceapi.ResourceSlice {
slice := testResourceSlice(goodName, goodName, driverName, 1)
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
slice.Finalizers = []string{
"example.com/foo",
}
return slice
}(),
},
"managed-fields": {
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
slice: func() *resourceapi.ResourceSlice {
slice := testResourceSlice(goodName, goodName, driverName, 1)
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
slice.ManagedFields = []metav1.ManagedFieldsEntry{
{
FieldsType: "FieldsV1",
Operation: "Apply",
APIVersion: "apps/v1",
Manager: "foo",
},
}
return slice
}(),
},
"good-labels": {
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
slice: func() *resourceapi.ResourceSlice {
slice := testResourceSlice(goodName, goodName, driverName, 1)
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
slice.Labels = map[string]string{
"apps.kubernetes.io/name": "test",
}
return slice
}(),
},
"bad-labels": {
wantFailures: field.ErrorList{field.Invalid(field.NewPath("metadata", "labels"), badValue, "a valid label must be an empty string or consist of alphanumeric characters, '-', '_' or '.', and must start and end with an alphanumeric character (e.g. 'MyValue', or 'my_value', or '12345', regex used for validation is '(([A-Za-z0-9][-A-Za-z0-9_.]*)?[A-Za-z0-9])?')")},
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
slice: func() *resourceapi.ResourceSlice {
slice := testResourceSlice(goodName, goodName, driverName, 1)
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
slice.Labels = map[string]string{
"hello-world": badValue,
}
return slice
}(),
},
"good-annotations": {
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
slice: func() *resourceapi.ResourceSlice {
slice := testResourceSlice(goodName, goodName, driverName, 1)
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
slice.Annotations = map[string]string{
"foo": "bar",
}
return slice
}(),
},
"bad-annotations": {
wantFailures: field.ErrorList{field.Invalid(field.NewPath("metadata", "annotations"), badName, "name part must consist of alphanumeric characters, '-', '_' or '.', and must start and end with an alphanumeric character (e.g. 'MyName', or 'my.name', or '123-abc', regex used for validation is '([A-Za-z0-9][-A-Za-z0-9_.]*)?[A-Za-z0-9]')")},
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
slice: func() *resourceapi.ResourceSlice {
slice := testResourceSlice(goodName, goodName, driverName, 1)
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
slice.Annotations = map[string]string{
badName: "hello world",
}
return slice
}(),
},
"bad-nodename": {
DRA: new API for 1.31 This is a complete revamp of the original API. Some of the key differences: - refocused on structured parameters and allocating devices - support for constraints across devices - support for allocating "all" or a fixed amount of similar devices in a single request - no class for ResourceClaims, instead individual device requests are associated with a mandatory DeviceClass For the sake of simplicity, optional basic types (ints, strings) where the null value is the default are represented as values in the API types. This makes Go code simpler because it doesn't have to check for nil (consumers) and values can be set directly (producers). The effect is that in protobuf, these fields always get encoded because `opt` only has an effect for pointers. The roundtrip test data for v1.29.0 and v1.30.0 changes because of the new "request" field. This is considered acceptable because the entire `claims` field in the pod spec is still alpha. The implementation is complete enough to bring up the apiserver. Adapting other components follows.
2024-06-18 23:47:29 +08:00
wantFailures: field.ErrorList{
field.Invalid(field.NewPath("spec", "pool", "name"), badName, "a lowercase RFC 1123 subdomain must consist of lower case alphanumeric characters, '-' or '.', and must start and end with an alphanumeric character (e.g. 'example.com', regex used for validation is '[a-z0-9]([-a-z0-9]*[a-z0-9])?(\\.[a-z0-9]([-a-z0-9]*[a-z0-9])?)*')"),
field.Invalid(field.NewPath("spec", "nodeName"), badName, "a lowercase RFC 1123 subdomain must consist of lower case alphanumeric characters, '-' or '.', and must start and end with an alphanumeric character (e.g. 'example.com', regex used for validation is '[a-z0-9]([-a-z0-9]*[a-z0-9])?(\\.[a-z0-9]([-a-z0-9]*[a-z0-9])?)*')"),
},
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
slice: testResourceSlice(goodName, badName, driverName, 1),
DRA: new API for 1.31 This is a complete revamp of the original API. Some of the key differences: - refocused on structured parameters and allocating devices - support for constraints across devices - support for allocating "all" or a fixed amount of similar devices in a single request - no class for ResourceClaims, instead individual device requests are associated with a mandatory DeviceClass For the sake of simplicity, optional basic types (ints, strings) where the null value is the default are represented as values in the API types. This makes Go code simpler because it doesn't have to check for nil (consumers) and values can be set directly (producers). The effect is that in protobuf, these fields always get encoded because `opt` only has an effect for pointers. The roundtrip test data for v1.29.0 and v1.30.0 changes because of the new "request" field. This is considered acceptable because the entire `claims` field in the pod spec is still alpha. The implementation is complete enough to bring up the apiserver. Adapting other components follows.
2024-06-18 23:47:29 +08:00
},
"bad-multi-pool-name": {
wantFailures: field.ErrorList{
field.Invalid(field.NewPath("spec", "pool", "name"), badName, "a lowercase RFC 1123 subdomain must consist of lower case alphanumeric characters, '-' or '.', and must start and end with an alphanumeric character (e.g. 'example.com', regex used for validation is '[a-z0-9]([-a-z0-9]*[a-z0-9])?(\\.[a-z0-9]([-a-z0-9]*[a-z0-9])?)*')"),
field.Invalid(field.NewPath("spec", "pool", "name"), badName, "a lowercase RFC 1123 subdomain must consist of lower case alphanumeric characters, '-' or '.', and must start and end with an alphanumeric character (e.g. 'example.com', regex used for validation is '[a-z0-9]([-a-z0-9]*[a-z0-9])?(\\.[a-z0-9]([-a-z0-9]*[a-z0-9])?)*')"),
field.Invalid(field.NewPath("spec", "nodeName"), badName+"/"+badName, "a lowercase RFC 1123 subdomain must consist of lower case alphanumeric characters, '-' or '.', and must start and end with an alphanumeric character (e.g. 'example.com', regex used for validation is '[a-z0-9]([-a-z0-9]*[a-z0-9])?(\\.[a-z0-9]([-a-z0-9]*[a-z0-9])?)*')"),
},
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
slice: testResourceSlice(goodName, badName+"/"+badName, driverName, 1),
},
"good-pool-name": {
slice: func() *resourceapi.ResourceSlice {
slice := testResourceSlice(goodName, goodName, driverName, 1)
slice.Spec.Pool.Name = strings.Repeat("x", resourceapi.PoolNameMaxLength)
return slice
}(),
},
"bad-pool": {
wantFailures: field.ErrorList{
field.TooLongMaxLength(field.NewPath("spec", "pool", "name"), strings.Repeat("x/", resourceapi.PoolNameMaxLength/2)+"xy", resourceapi.PoolNameMaxLength),
field.Invalid(field.NewPath("spec", "pool", "resourceSliceCount"), int64(0), "must be greater than zero"),
field.Invalid(field.NewPath("spec", "pool", "generation"), int64(-1), "must be greater than or equal to zero"),
},
slice: func() *resourceapi.ResourceSlice {
slice := testResourceSlice(goodName, goodName, driverName, 1)
slice.Spec.Pool.Name = strings.Repeat("x/", resourceapi.PoolNameMaxLength/2) + "xy"
slice.Spec.Pool.ResourceSliceCount = 0
slice.Spec.Pool.Generation = -1
return slice
}(),
},
"missing-pool-name": {
wantFailures: field.ErrorList{
field.Required(field.NewPath("spec", "pool", "name"), ""),
},
slice: func() *resourceapi.ResourceSlice {
slice := testResourceSlice(goodName, goodName, driverName, 1)
slice.Spec.Pool.Name = ""
return slice
}(),
},
"bad-empty-node-selector": {
wantFailures: field.ErrorList{
field.Required(field.NewPath("spec", "nodeSelector", "nodeSelectorTerms"), "must have at least one node selector term"), // From core validation.
field.Invalid(field.NewPath("spec", "nodeSelector", "nodeSelectorTerms"), []core.NodeSelectorTerm(nil), "must have exactly one node selector term"), // From DRA validation.
},
slice: func() *resourceapi.ResourceSlice {
slice := testResourceSlice(goodName, goodName, driverName, 1)
slice.Spec.NodeName = ""
slice.Spec.NodeSelector = &core.NodeSelector{}
return slice
}(),
},
"bad-node-selection": {
wantFailures: field.ErrorList{field.Invalid(field.NewPath("spec"), nil, "exactly one of `nodeName`, `nodeSelector`, or `allNodes` is required")},
slice: func() *resourceapi.ResourceSlice {
slice := testResourceSlice(goodName, goodName, driverName, 1)
slice.Spec.NodeName = "worker"
slice.Spec.NodeSelector = &core.NodeSelector{
NodeSelectorTerms: []core.NodeSelectorTerm{{MatchFields: []core.NodeSelectorRequirement{{Key: "metadata.name", Operator: core.NodeSelectorOpIn, Values: []string{"worker"}}}}},
}
return slice
}(),
},
"bad-node-selection-all-nodes": {
wantFailures: field.ErrorList{field.Invalid(field.NewPath("spec"), nil, "exactly one of `nodeName`, `nodeSelector`, or `allNodes` is required")},
slice: func() *resourceapi.ResourceSlice {
slice := testResourceSlice(goodName, goodName, driverName, 1)
slice.Spec.NodeName = "worker"
slice.Spec.AllNodes = true
return slice
}(),
},
"empty-node-selection": {
wantFailures: field.ErrorList{field.Required(field.NewPath("spec"), "exactly one of `nodeName`, `nodeSelector`, or `allNodes` is required")},
slice: func() *resourceapi.ResourceSlice {
slice := testResourceSlice(goodName, goodName, driverName, 1)
slice.Spec.NodeName = ""
return slice
}(),
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
},
"bad-drivername": {
DRA: new API for 1.31 This is a complete revamp of the original API. Some of the key differences: - refocused on structured parameters and allocating devices - support for constraints across devices - support for allocating "all" or a fixed amount of similar devices in a single request - no class for ResourceClaims, instead individual device requests are associated with a mandatory DeviceClass For the sake of simplicity, optional basic types (ints, strings) where the null value is the default are represented as values in the API types. This makes Go code simpler because it doesn't have to check for nil (consumers) and values can be set directly (producers). The effect is that in protobuf, these fields always get encoded because `opt` only has an effect for pointers. The roundtrip test data for v1.29.0 and v1.30.0 changes because of the new "request" field. This is considered acceptable because the entire `claims` field in the pod spec is still alpha. The implementation is complete enough to bring up the apiserver. Adapting other components follows.
2024-06-18 23:47:29 +08:00
wantFailures: field.ErrorList{field.Invalid(field.NewPath("spec", "driver"), badName, "a lowercase RFC 1123 subdomain must consist of lower case alphanumeric characters, '-' or '.', and must start and end with an alphanumeric character (e.g. 'example.com', regex used for validation is '[a-z0-9]([-a-z0-9]*[a-z0-9])?(\\.[a-z0-9]([-a-z0-9]*[a-z0-9])?)*')")},
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
slice: testResourceSlice(goodName, goodName, badName, 1),
},
"bad-devices": {
wantFailures: field.ErrorList{
field.Invalid(field.NewPath("spec", "devices").Index(1).Child("name"), badName, "a lowercase RFC 1123 label must consist of lower case alphanumeric characters or '-', and must start and end with an alphanumeric character (e.g. 'my-name', or '123-abc', regex used for validation is '[a-z0-9]([-a-z0-9]*[a-z0-9])?')"),
field.Required(field.NewPath("spec", "devices").Index(2).Child("basic"), ""),
},
slice: func() *resourceapi.ResourceSlice {
slice := testResourceSlice(goodName, goodName, goodName, 3)
slice.Spec.Devices[1].Name = badName
slice.Spec.Devices[2].Basic = nil
return slice
}(),
},
"bad-attribute": {
wantFailures: field.ErrorList{
field.TypeInvalid(field.NewPath("spec", "devices").Index(1).Child("basic", "attributes").Key(badName), badName, "a valid C identifier must start with alphabetic character or '_', followed by a string of alphanumeric characters or '_' (e.g. 'my_name', or 'MY_NAME', or 'MyName', regex used for validation is '[A-Za-z_][A-Za-z0-9_]*')"),
field.Required(field.NewPath("spec", "devices").Index(1).Child("basic", "attributes").Key(badName), "exactly one value must be specified"),
field.Invalid(field.NewPath("spec", "devices").Index(2).Child("basic", "attributes").Key(goodName), resourceapi.DeviceAttribute{StringValue: ptr.To("x"), VersionValue: ptr.To("1.2.3")}, "exactly one value must be specified"),
field.Invalid(field.NewPath("spec", "devices").Index(3).Child("basic", "attributes").Key(goodName).Child("version"), strings.Repeat("x", resourceapi.DeviceAttributeMaxValueLength+1), "must be a string compatible with semver.org spec 2.0.0"),
field.TooLongMaxLength(field.NewPath("spec", "devices").Index(3).Child("basic", "attributes").Key(goodName).Child("version"), strings.Repeat("x", resourceapi.DeviceAttributeMaxValueLength+1), resourceapi.DeviceAttributeMaxValueLength),
field.TooLongMaxLength(field.NewPath("spec", "devices").Index(4).Child("basic", "attributes").Key(goodName).Child("string"), strings.Repeat("x", resourceapi.DeviceAttributeMaxValueLength+1), resourceapi.DeviceAttributeMaxValueLength),
},
slice: func() *resourceapi.ResourceSlice {
slice := testResourceSlice(goodName, goodName, goodName, 5)
slice.Spec.Devices[1].Basic.Attributes = map[resourceapi.QualifiedName]resourceapi.DeviceAttribute{
resourceapi.QualifiedName(badName): {},
}
slice.Spec.Devices[2].Basic.Attributes = map[resourceapi.QualifiedName]resourceapi.DeviceAttribute{
resourceapi.QualifiedName(goodName): {StringValue: ptr.To("x"), VersionValue: ptr.To("1.2.3")},
}
slice.Spec.Devices[3].Basic.Attributes = map[resourceapi.QualifiedName]resourceapi.DeviceAttribute{
resourceapi.QualifiedName(goodName): {VersionValue: ptr.To(strings.Repeat("x", resourceapi.DeviceAttributeMaxValueLength+1))},
}
slice.Spec.Devices[4].Basic.Attributes = map[resourceapi.QualifiedName]resourceapi.DeviceAttribute{
resourceapi.QualifiedName(goodName): {StringValue: ptr.To(strings.Repeat("x", resourceapi.DeviceAttributeMaxValueLength+1))},
}
return slice
}(),
},
"good-attribute-names": {
slice: func() *resourceapi.ResourceSlice {
slice := testResourceSlice(goodName, goodName, goodName, 2)
slice.Spec.Devices[1].Basic.Attributes = map[resourceapi.QualifiedName]resourceapi.DeviceAttribute{
resourceapi.QualifiedName(strings.Repeat("x", resourceapi.DeviceMaxIDLength)): {StringValue: ptr.To("y")},
resourceapi.QualifiedName(strings.Repeat("x", resourceapi.DeviceMaxDomainLength) + "/" + strings.Repeat("y", resourceapi.DeviceMaxIDLength)): {StringValue: ptr.To("z")},
}
return slice
}(),
},
"bad-attribute-c-identifier": {
wantFailures: field.ErrorList{
field.TooLongMaxLength(field.NewPath("spec", "devices").Index(1).Child("basic", "attributes").Key(strings.Repeat(".", resourceapi.DeviceMaxIDLength+1)), strings.Repeat(".", resourceapi.DeviceMaxIDLength+1), resourceapi.DeviceMaxIDLength),
field.TypeInvalid(field.NewPath("spec", "devices").Index(1).Child("basic", "attributes").Key(strings.Repeat(".", resourceapi.DeviceMaxIDLength+1)), strings.Repeat(".", resourceapi.DeviceMaxIDLength+1), "a valid C identifier must start with alphabetic character or '_', followed by a string of alphanumeric characters or '_' (e.g. 'my_name', or 'MY_NAME', or 'MyName', regex used for validation is '[A-Za-z_][A-Za-z0-9_]*')"),
},
slice: func() *resourceapi.ResourceSlice {
slice := testResourceSlice(goodName, goodName, goodName, 2)
slice.Spec.Devices[1].Basic.Attributes = map[resourceapi.QualifiedName]resourceapi.DeviceAttribute{
resourceapi.QualifiedName(strings.Repeat(".", resourceapi.DeviceMaxIDLength+1)): {StringValue: ptr.To("y")},
}
return slice
}(),
},
"bad-attribute-domain": {
wantFailures: field.ErrorList{
field.TooLong(field.NewPath("spec", "devices").Index(1).Child("basic", "attributes").Key(strings.Repeat("_", resourceapi.DeviceMaxDomainLength+1)+"/y"), strings.Repeat("_", resourceapi.DeviceMaxDomainLength+1), resourceapi.DeviceMaxDomainLength),
field.Invalid(field.NewPath("spec", "devices").Index(1).Child("basic", "attributes").Key(strings.Repeat("_", resourceapi.DeviceMaxDomainLength+1)+"/y"), strings.Repeat("_", resourceapi.DeviceMaxDomainLength+1), "a lowercase RFC 1123 subdomain must consist of lower case alphanumeric characters, '-' or '.', and must start and end with an alphanumeric character (e.g. 'example.com', regex used for validation is '[a-z0-9]([-a-z0-9]*[a-z0-9])?(\\.[a-z0-9]([-a-z0-9]*[a-z0-9])?)*')"),
},
slice: func() *resourceapi.ResourceSlice {
slice := testResourceSlice(goodName, goodName, goodName, 2)
slice.Spec.Devices[1].Basic.Attributes = map[resourceapi.QualifiedName]resourceapi.DeviceAttribute{
resourceapi.QualifiedName(strings.Repeat("_", resourceapi.DeviceMaxDomainLength+1) + "/y"): {StringValue: ptr.To("z")},
}
return slice
}(),
},
"bad-key-too-long": {
wantFailures: field.ErrorList{
field.TooLong(field.NewPath("spec", "devices").Index(1).Child("basic", "attributes").Key("xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx...xxxxxxxxxxxx/yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy"), strings.Repeat("x", resourceapi.DeviceMaxDomainLength+1), resourceapi.DeviceMaxDomainLength),
field.TooLongMaxLength(field.NewPath("spec", "devices").Index(1).Child("basic", "attributes").Key("xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx...xxxxxxxxxxxx/yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy"), strings.Repeat("y", resourceapi.DeviceMaxIDLength+1), resourceapi.DeviceMaxIDLength),
},
slice: func() *resourceapi.ResourceSlice {
slice := testResourceSlice(goodName, goodName, goodName, 2)
slice.Spec.Devices[1].Basic.Attributes = map[resourceapi.QualifiedName]resourceapi.DeviceAttribute{
resourceapi.QualifiedName(strings.Repeat("x", resourceapi.DeviceMaxDomainLength+1) + "/" + strings.Repeat("y", resourceapi.DeviceMaxIDLength+1)): {StringValue: ptr.To("z")},
}
return slice
}(),
},
"bad-attribute-empty-domain-and-c-identifier": {
wantFailures: field.ErrorList{
field.Required(field.NewPath("spec", "devices").Index(1).Child("basic", "attributes").Key("/"), "the domain must not be empty"),
field.Required(field.NewPath("spec", "devices").Index(1).Child("basic", "attributes").Key("/"), "the name must not be empty"),
},
slice: func() *resourceapi.ResourceSlice {
slice := testResourceSlice(goodName, goodName, goodName, 2)
slice.Spec.Devices[1].Basic.Attributes = map[resourceapi.QualifiedName]resourceapi.DeviceAttribute{
resourceapi.QualifiedName("/"): {StringValue: ptr.To("z")},
}
return slice
}(),
},
"combined-attributes-and-capacity-length": {
wantFailures: field.ErrorList{
field.Invalid(field.NewPath("spec", "devices").Index(2).Child("basic"), resourceapi.ResourceSliceMaxAttributesAndCapacitiesPerDevice+1, fmt.Sprintf("the total number of attributes and capacities must not exceed %d", resourceapi.ResourceSliceMaxAttributesAndCapacitiesPerDevice)),
},
slice: func() *resourceapi.ResourceSlice {
slice := testResourceSlice(goodName, goodName, goodName, 3)
slice.Spec.Devices[0].Basic.Attributes = map[resourceapi.QualifiedName]resourceapi.DeviceAttribute{}
DRA API: use DeviceCapacity struct instead of plain Quantity This enables a future extension where capacity of a single device gets consumed by different claims. The semantic without any additional fields is the same as before: a capacity cannot be split up and is only an attribute of a device. Because its semantically the same as before, two-way conversion to v1alpha3 is possible.
2024-09-26 22:56:48 +08:00
slice.Spec.Devices[0].Basic.Capacity = map[resourceapi.QualifiedName]resourceapi.DeviceCapacity{}
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
for i := 0; i < resourceapi.ResourceSliceMaxAttributesAndCapacitiesPerDevice; i++ {
slice.Spec.Devices[0].Basic.Attributes[resourceapi.QualifiedName(fmt.Sprintf("attr_%d", i))] = resourceapi.DeviceAttribute{StringValue: ptr.To("x")}
}
slice.Spec.Devices[1].Basic.Attributes = map[resourceapi.QualifiedName]resourceapi.DeviceAttribute{}
DRA API: use DeviceCapacity struct instead of plain Quantity This enables a future extension where capacity of a single device gets consumed by different claims. The semantic without any additional fields is the same as before: a capacity cannot be split up and is only an attribute of a device. Because its semantically the same as before, two-way conversion to v1alpha3 is possible.
2024-09-26 22:56:48 +08:00
slice.Spec.Devices[1].Basic.Capacity = map[resourceapi.QualifiedName]resourceapi.DeviceCapacity{}
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
quantity := resource.MustParse("1Gi")
DRA API: rename DeviceCapacity.Quantity to DeviceCapacity.Value Based on review feedback (https://github.com/kubernetes/kubernetes/pull/127511#discussion_r1823521172).
2024-11-01 04:30:28 +08:00
capacity := resourceapi.DeviceCapacity{Value: quantity}
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
for i := 0; i < resourceapi.ResourceSliceMaxAttributesAndCapacitiesPerDevice; i++ {
DRA API: use DeviceCapacity struct instead of plain Quantity This enables a future extension where capacity of a single device gets consumed by different claims. The semantic without any additional fields is the same as before: a capacity cannot be split up and is only an attribute of a device. Because its semantically the same as before, two-way conversion to v1alpha3 is possible.
2024-09-26 22:56:48 +08:00
slice.Spec.Devices[1].Basic.Capacity[resourceapi.QualifiedName(fmt.Sprintf("cap_%d", i))] = capacity
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
}
// Too large together by one.
slice.Spec.Devices[2].Basic.Attributes = slice.Spec.Devices[0].Basic.Attributes
DRA API: use DeviceCapacity struct instead of plain Quantity This enables a future extension where capacity of a single device gets consumed by different claims. The semantic without any additional fields is the same as before: a capacity cannot be split up and is only an attribute of a device. Because its semantically the same as before, two-way conversion to v1alpha3 is possible.
2024-09-26 22:56:48 +08:00
slice.Spec.Devices[2].Basic.Capacity = map[resourceapi.QualifiedName]resourceapi.DeviceCapacity{
"cap": capacity,
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
}
return slice
}(),
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
},
DRA API: validate node selector labels Previously, ValidateNodeSelector did not check that labels are valid. Now it does for resource.k8s.io, regardless whether an object already was created with invalid labels in an earlier Kubernetes release. Theoretically this is a breaking change and could cause problems during an upgrade, but that is highly unlikely in practice. In contrast to node affinity, DRA does not ignore parse errors (= uses NewNodeSelector, not NewLazyErrorNodeSelector), so invalid labels would have been found instead of being silently ignored. Even if some object has invalid labels, this only affects an alpha -> beta upgrade which isn't guaranteed to work seamlessly.
2024-10-23 17:28:27 +08:00
"invalid-node-selecor-label-value": {
wantFailures: field.ErrorList{field.Invalid(field.NewPath("spec", "nodeSelector", "nodeSelectorTerms").Index(0).Child("matchExpressions").Index(0).Child("values").Index(0), "-1", "a valid label must be an empty string or consist of alphanumeric characters, '-', '_' or '.', and must start and end with an alphanumeric character (e.g. 'MyValue', or 'my_value', or '12345', regex used for validation is '(([A-Za-z0-9][-A-Za-z0-9_.]*)?[A-Za-z0-9])?')")},
slice: func() *resourceapi.ResourceSlice {
slice := testResourceSlice(goodName, goodName, goodName, 3)
slice.Spec.NodeName = ""
slice.Spec.NodeSelector = &core.NodeSelector{
NodeSelectorTerms: []core.NodeSelectorTerm{{
MatchExpressions: []core.NodeSelectorRequirement{{
Key: "foo",
Operator: core.NodeSelectorOpIn,
Values: []string{"-1"},
}},
}},
}
return slice
}(),
},
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
}
for name, scenario := range scenarios {
t.Run(name, func(t *testing.T) {
dra api: rename NodeResourceSlice -> ResourceSlice While currently those objects only get published by the kubelet for node-local resources, this could change once we also support network-attached resources. Dropping the "Node" prefix enables such a future extension. The NodeName in ResourceSlice and StructuredResourceHandle then becomes optional. The kubelet still needs to provide one and it must match its own node name, otherwise it doesn't have permission to access ResourceSlice objects.
2024-03-07 17:14:11 +08:00
errs := ValidateResourceSlice(scenario.slice)
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
assertFailures(t, scenario.wantFailures, errs)
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
})
}
}
dra api: rename NodeResourceSlice -> ResourceSlice While currently those objects only get published by the kubelet for node-local resources, this could change once we also support network-attached resources. Dropping the "Node" prefix enables such a future extension. The NodeName in ResourceSlice and StructuredResourceHandle then becomes optional. The kubelet still needs to provide one and it must match its own node name, otherwise it doesn't have permission to access ResourceSlice objects.
2024-03-07 17:14:11 +08:00
func TestValidateResourceSliceUpdate(t *testing.T) {
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
name := "valid"
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
validResourceSlice := testResourceSlice(name, name, name, 1)
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
scenarios := map[string]struct {
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
oldResourceSlice *resourceapi.ResourceSlice
update func(slice *resourceapi.ResourceSlice) *resourceapi.ResourceSlice
dra api: rename NodeResourceSlice -> ResourceSlice While currently those objects only get published by the kubelet for node-local resources, this could change once we also support network-attached resources. Dropping the "Node" prefix enables such a future extension. The NodeName in ResourceSlice and StructuredResourceHandle then becomes optional. The kubelet still needs to provide one and it must match its own node name, otherwise it doesn't have permission to access ResourceSlice objects.
2024-03-07 17:14:11 +08:00
wantFailures field.ErrorList
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
}{
"valid-no-op-update": {
dra api: rename NodeResourceSlice -> ResourceSlice While currently those objects only get published by the kubelet for node-local resources, this could change once we also support network-attached resources. Dropping the "Node" prefix enables such a future extension. The NodeName in ResourceSlice and StructuredResourceHandle then becomes optional. The kubelet still needs to provide one and it must match its own node name, otherwise it doesn't have permission to access ResourceSlice objects.
2024-03-07 17:14:11 +08:00
oldResourceSlice: validResourceSlice,
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
update: func(slice *resourceapi.ResourceSlice) *resourceapi.ResourceSlice { return slice },
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
},
"invalid-name-update": {
dra api: rename NodeResourceSlice -> ResourceSlice While currently those objects only get published by the kubelet for node-local resources, this could change once we also support network-attached resources. Dropping the "Node" prefix enables such a future extension. The NodeName in ResourceSlice and StructuredResourceHandle then becomes optional. The kubelet still needs to provide one and it must match its own node name, otherwise it doesn't have permission to access ResourceSlice objects.
2024-03-07 17:14:11 +08:00
oldResourceSlice: validResourceSlice,
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
update: func(slice *resourceapi.ResourceSlice) *resourceapi.ResourceSlice {
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
slice.Name += "-update"
return slice
},
wantFailures: field.ErrorList{field.Invalid(field.NewPath("metadata", "name"), name+"-update", "field is immutable")},
},
"invalid-update-nodename": {
DRA: new API for 1.31 This is a complete revamp of the original API. Some of the key differences: - refocused on structured parameters and allocating devices - support for constraints across devices - support for allocating "all" or a fixed amount of similar devices in a single request - no class for ResourceClaims, instead individual device requests are associated with a mandatory DeviceClass For the sake of simplicity, optional basic types (ints, strings) where the null value is the default are represented as values in the API types. This makes Go code simpler because it doesn't have to check for nil (consumers) and values can be set directly (producers). The effect is that in protobuf, these fields always get encoded because `opt` only has an effect for pointers. The roundtrip test data for v1.29.0 and v1.30.0 changes because of the new "request" field. This is considered acceptable because the entire `claims` field in the pod spec is still alpha. The implementation is complete enough to bring up the apiserver. Adapting other components follows.
2024-06-18 23:47:29 +08:00
wantFailures: field.ErrorList{field.Invalid(field.NewPath("spec", "nodeName"), name+"-updated", "field is immutable")},
dra api: rename NodeResourceSlice -> ResourceSlice While currently those objects only get published by the kubelet for node-local resources, this could change once we also support network-attached resources. Dropping the "Node" prefix enables such a future extension. The NodeName in ResourceSlice and StructuredResourceHandle then becomes optional. The kubelet still needs to provide one and it must match its own node name, otherwise it doesn't have permission to access ResourceSlice objects.
2024-03-07 17:14:11 +08:00
oldResourceSlice: validResourceSlice,
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
update: func(slice *resourceapi.ResourceSlice) *resourceapi.ResourceSlice {
DRA: new API for 1.31 This is a complete revamp of the original API. Some of the key differences: - refocused on structured parameters and allocating devices - support for constraints across devices - support for allocating "all" or a fixed amount of similar devices in a single request - no class for ResourceClaims, instead individual device requests are associated with a mandatory DeviceClass For the sake of simplicity, optional basic types (ints, strings) where the null value is the default are represented as values in the API types. This makes Go code simpler because it doesn't have to check for nil (consumers) and values can be set directly (producers). The effect is that in protobuf, these fields always get encoded because `opt` only has an effect for pointers. The roundtrip test data for v1.29.0 and v1.30.0 changes because of the new "request" field. This is considered acceptable because the entire `claims` field in the pod spec is still alpha. The implementation is complete enough to bring up the apiserver. Adapting other components follows.
2024-06-18 23:47:29 +08:00
slice.Spec.NodeName += "-updated"
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
return slice
},
},
"invalid-update-drivername": {
DRA: new API for 1.31 This is a complete revamp of the original API. Some of the key differences: - refocused on structured parameters and allocating devices - support for constraints across devices - support for allocating "all" or a fixed amount of similar devices in a single request - no class for ResourceClaims, instead individual device requests are associated with a mandatory DeviceClass For the sake of simplicity, optional basic types (ints, strings) where the null value is the default are represented as values in the API types. This makes Go code simpler because it doesn't have to check for nil (consumers) and values can be set directly (producers). The effect is that in protobuf, these fields always get encoded because `opt` only has an effect for pointers. The roundtrip test data for v1.29.0 and v1.30.0 changes because of the new "request" field. This is considered acceptable because the entire `claims` field in the pod spec is still alpha. The implementation is complete enough to bring up the apiserver. Adapting other components follows.
2024-06-18 23:47:29 +08:00
wantFailures: field.ErrorList{field.Invalid(field.NewPath("spec", "driver"), name+"-updated", "field is immutable")},
oldResourceSlice: validResourceSlice,
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
update: func(slice *resourceapi.ResourceSlice) *resourceapi.ResourceSlice {
DRA: new API for 1.31 This is a complete revamp of the original API. Some of the key differences: - refocused on structured parameters and allocating devices - support for constraints across devices - support for allocating "all" or a fixed amount of similar devices in a single request - no class for ResourceClaims, instead individual device requests are associated with a mandatory DeviceClass For the sake of simplicity, optional basic types (ints, strings) where the null value is the default are represented as values in the API types. This makes Go code simpler because it doesn't have to check for nil (consumers) and values can be set directly (producers). The effect is that in protobuf, these fields always get encoded because `opt` only has an effect for pointers. The roundtrip test data for v1.29.0 and v1.30.0 changes because of the new "request" field. This is considered acceptable because the entire `claims` field in the pod spec is still alpha. The implementation is complete enough to bring up the apiserver. Adapting other components follows.
2024-06-18 23:47:29 +08:00
slice.Spec.Driver += "-updated"
return slice
},
},
"invalid-update-pool": {
wantFailures: field.ErrorList{field.Invalid(field.NewPath("spec", "pool", "name"), validResourceSlice.Spec.Pool.Name+"-updated", "field is immutable")},
dra api: rename NodeResourceSlice -> ResourceSlice While currently those objects only get published by the kubelet for node-local resources, this could change once we also support network-attached resources. Dropping the "Node" prefix enables such a future extension. The NodeName in ResourceSlice and StructuredResourceHandle then becomes optional. The kubelet still needs to provide one and it must match its own node name, otherwise it doesn't have permission to access ResourceSlice objects.
2024-03-07 17:14:11 +08:00
oldResourceSlice: validResourceSlice,
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
update: func(slice *resourceapi.ResourceSlice) *resourceapi.ResourceSlice {
DRA: new API for 1.31 This is a complete revamp of the original API. Some of the key differences: - refocused on structured parameters and allocating devices - support for constraints across devices - support for allocating "all" or a fixed amount of similar devices in a single request - no class for ResourceClaims, instead individual device requests are associated with a mandatory DeviceClass For the sake of simplicity, optional basic types (ints, strings) where the null value is the default are represented as values in the API types. This makes Go code simpler because it doesn't have to check for nil (consumers) and values can be set directly (producers). The effect is that in protobuf, these fields always get encoded because `opt` only has an effect for pointers. The roundtrip test data for v1.29.0 and v1.30.0 changes because of the new "request" field. This is considered acceptable because the entire `claims` field in the pod spec is still alpha. The implementation is complete enough to bring up the apiserver. Adapting other components follows.
2024-06-18 23:47:29 +08:00
slice.Spec.Pool.Name += "-updated"
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
return slice
},
},
DRA API: validate node selector labels Previously, ValidateNodeSelector did not check that labels are valid. Now it does for resource.k8s.io, regardless whether an object already was created with invalid labels in an earlier Kubernetes release. Theoretically this is a breaking change and could cause problems during an upgrade, but that is highly unlikely in practice. In contrast to node affinity, DRA does not ignore parse errors (= uses NewNodeSelector, not NewLazyErrorNodeSelector), so invalid labels would have been found instead of being silently ignored. Even if some object has invalid labels, this only affects an alpha -> beta upgrade which isn't guaranteed to work seamlessly.
2024-10-23 17:28:27 +08:00
"invalid-update-to-invalid-nodeselector-label-value": {
wantFailures: field.ErrorList{field.Invalid(field.NewPath("spec", "nodeSelector", "nodeSelectorTerms").Index(0).Child("matchExpressions").Index(0).Child("values").Index(0), "-1", "a valid label must be an empty string or consist of alphanumeric characters, '-', '_' or '.', and must start and end with an alphanumeric character (e.g. 'MyValue', or 'my_value', or '12345', regex used for validation is '(([A-Za-z0-9][-A-Za-z0-9_.]*)?[A-Za-z0-9])?')")},
oldResourceSlice: func() *resourceapi.ResourceSlice {
slice := validResourceSlice.DeepCopy()
slice.Spec.NodeName = ""
slice.Spec.NodeSelector = &core.NodeSelector{
NodeSelectorTerms: []core.NodeSelectorTerm{{
MatchExpressions: []core.NodeSelectorRequirement{{
Key: "foo",
Operator: core.NodeSelectorOpIn,
Values: []string{"bar"},
}},
}},
}
return slice
}(),
update: func(slice *resourceapi.ResourceSlice) *resourceapi.ResourceSlice {
slice.Spec.NodeSelector = &core.NodeSelector{
NodeSelectorTerms: []core.NodeSelectorTerm{{
MatchExpressions: []core.NodeSelectorRequirement{{
Key: "foo",
Operator: core.NodeSelectorOpIn,
Values: []string{"-1"},
}},
}},
}
return slice
},
},
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
}
for name, scenario := range scenarios {
t.Run(name, func(t *testing.T) {
dra api: rename NodeResourceSlice -> ResourceSlice While currently those objects only get published by the kubelet for node-local resources, this could change once we also support network-attached resources. Dropping the "Node" prefix enables such a future extension. The NodeName in ResourceSlice and StructuredResourceHandle then becomes optional. The kubelet still needs to provide one and it must match its own node name, otherwise it doesn't have permission to access ResourceSlice objects.
2024-03-07 17:14:11 +08:00
scenario.oldResourceSlice.ResourceVersion = "1"
errs := ValidateResourceSliceUpdate(scenario.update(scenario.oldResourceSlice.DeepCopy()), scenario.oldResourceSlice)
DRA API: enhance validation testing The line coverage is now at 98.5% and several more corner cases are covered. The remaining lines are hard or impossible to reach. The actual validation is the same as before, with some small tweaks to the generated errors. When failures are not as expected, it is useful to show what the expected and actual failures look like to a user. Perhaps even better would be to put the expected texts into the test files instead of the error structs. That would be easier to review and shorter.
2024-09-25 00:52:24 +08:00
assertFailures(t, scenario.wantFailures, errs)
dra api: add structured parameters NodeResourceSlice will be used by kubelet to publish resource information on behalf of DRA drivers on the node. NodeName and DriverName in NodeResourceSlice must be immutable. This simplifies tracking the different objects because what they are for cannot change after creation. The new field in ResourceClass tells scheduler and autoscaler that they are expected to handle allocation. ResourceClaimParameters and ResourceClassParameters are new types for telling in-tree components how to handle claims.
2024-02-14 21:38:42 +08:00
})
}
}