cherry pick main

.github/workflows/ci.yml

@@ -25,3 +25,6 @@ jobs:
         run: |
           make fmt && git add pkg &&
           git diff --cached --exit-code || (echo 'Please run "make fmt" to verify gofmt' && exit 1);
+      - name: Check inline fields
+        run: |
+          go run cmd/inlinecheck/main.go

cmd/inlinecheck/main.go

@@ -0,0 +1,106 @@
+package main
+
+import (
+	"fmt"
+	"os"
+	"reflect"
+	"strings"
+
+	autoscalingv1a1 "github.com/kubewharf/katalyst-api/pkg/apis/autoscaling/v1alpha1"
+	autoscalingv1a2 "github.com/kubewharf/katalyst-api/pkg/apis/autoscaling/v1alpha2"
+	configv1a1 "github.com/kubewharf/katalyst-api/pkg/apis/config/v1alpha1"
+	nodev1a1 "github.com/kubewharf/katalyst-api/pkg/apis/node/v1alpha1"
+	overcommitv1a1 "github.com/kubewharf/katalyst-api/pkg/apis/overcommit/v1alpha1"
+	recommendationv1a1 "github.com/kubewharf/katalyst-api/pkg/apis/recommendation/v1alpha1"
+	schedcfgv1b3 "github.com/kubewharf/katalyst-api/pkg/apis/scheduling/config/v1beta3"
+	tidev1a1 "github.com/kubewharf/katalyst-api/pkg/apis/tide/v1alpha1"
+	workloadv1a1 "github.com/kubewharf/katalyst-api/pkg/apis/workload/v1alpha1"
+
+	"k8s.io/apimachinery/pkg/runtime"
+)
+
+func main() {
+	scheme := runtime.NewScheme()
+	configv1a1.AddToScheme(scheme)
+	autoscalingv1a1.AddToScheme(scheme)
+	autoscalingv1a2.AddToScheme(scheme)
+	nodev1a1.AddToScheme(scheme)
+	overcommitv1a1.AddToScheme(scheme)
+	recommendationv1a1.AddToScheme(scheme)
+	schedcfgv1b3.AddToScheme(scheme)
+	tidev1a1.AddToScheme(scheme)
+	workloadv1a1.AddToScheme(scheme)
+
+	seenTypes := make(map[reflect.Type]struct{})
+	var errs []error
+	for _, typ := range scheme.AllKnownTypes() {
+		fmt.Printf("Checking %s.%s\n", typ.PkgPath(), typ.Name())
+		checkType(typ, typ.Name(), seenTypes, &errs)
+	}
+
+	for _, err := range errs {
+		fmt.Println(err)
+	}
+
+	if len(errs) > 0 {
+		os.Exit(1)
+	}
+}
+
+func parseTag(tag string) (name string) {
+	if idx := strings.Index(tag, ","); idx != -1 {
+		return tag[:idx]
+	} else {
+		return tag
+	}
+}
+
+// typ must be a struct type
+func checkType(typ reflect.Type, path string, seenTypes map[reflect.Type]struct{}, errs *[]error) {
+	if _, ok := seenTypes[typ]; ok {
+		return
+	}
+
+	for i := 0; i < typ.NumField(); i++ {
+		field := typ.Field(i)
+		if !field.IsExported() {
+			continue
+		}
+		fieldTyp := field.Type
+		origFieldTyp := fieldTyp
+
+		if fieldTyp.Kind() == reflect.Ptr ||
+			fieldTyp.Kind() == reflect.Slice ||
+			fieldTyp.Kind() == reflect.Array ||
+			fieldTyp.Kind() == reflect.Map {
+			fieldTyp = fieldTyp.Elem()
+		}
+		if fieldTyp.Kind() != reflect.Struct {
+			continue
+		}
+
+		var newPath string
+		switch origFieldTyp.Kind() {
+		case reflect.Struct, reflect.Ptr:
+			newPath = fmt.Sprintf("%s.%s", path, field.Name)
+		case reflect.Array, reflect.Slice:
+			newPath = fmt.Sprintf("%s.%s[0]", path, field.Name)
+		case reflect.Map:
+			newPath = fmt.Sprintf("%s.%s[*]", path, field.Name)
+		default:
+			continue
+		}
+
+		tag := field.Tag.Get("json")
+		name := parseTag(tag)
+		if name == "" && !field.Anonymous {
+			*errs = append(
+				*errs,
+				fmt.Errorf("field %s has no json tag and is not embedded, will cause unpredictable deserialization", newPath),
+			)
+		}
+		checkType(fieldTyp, newPath, seenTypes, errs)
+	}
+
+	seenTypes[typ] = struct{}{}
+}

config/crd/bases/config.katalyst.kubewharf.io_adminqosconfigurations.yaml

@@ -284,6 +284,58 @@ spec:
                                 minimum: 0
                                 type: number
                             type: object
+                          numaSysCPUPressureEvictionConfig:
+                            description: NumaSysCPUPressureEvictionConfig holds configurations
+                              for NUMA-level system CPU pressure eviction.
+                            properties:
+                              enableEviction:
+                                description: EnableEviction indicates whether to enable
+                                  NUMA-level system CPU pressure eviction.
+                                type: boolean
+                              gracePeriod:
+                                description: |-
+                                  GracePeriod is the grace period (in seconds) after a pod starts before it can be considered for eviction
+                                  due to NUMA system CPU pressure. 0 means no grace period.
+                                format: int64
+                                type: integer
+                              metricRingSize:
+                                description: MetricRingSize is the size of the metric
+                                  ring buffer for calculating NUMA system CPU pressure.
+                                type: integer
+                              numaCPUUsageHardThreshold:
+                                description: NumaCPUUsageHardThreshold is the hard
+                                  threshold for NUMA system CPU pressure.
+                                type: number
+                              numaCPUUsageSoftThreshold:
+                                description: NumaCPUUsageSoftThreshold is the soft
+                                  threshold for NUMA system CPU pressure.
+                                type: number
+                              numaSysOverTotalUsageEvictionThreshold:
+                                description: NUMASysOverTotalUsageEvictionThreshold
+                                  is the eviction threshold for NUMA system CPU pressure
+                                  over total system CPU pressure.
+                                type: number
+                              numaSysOverTotalUsageHardThreshold:
+                                description: NUMASysOverTotalUsageHardThreshold is
+                                  the hard threshold for NUMA system CPU pressure
+                                  over total system CPU pressure.
+                                type: number
+                              numaSysOverTotalUsageSoftThreshold:
+                                description: NUMASysOverTotalUsageSoftThreshold is
+                                  the soft threshold for NUMA system CPU pressure
+                                  over total system CPU pressure.
+                                type: number
+                              syncPeriod:
+                                description: SyncPeriod is the sync period (in seconds)
+                                  for updating NUMA system CPU pressure metrics.
+                                format: int64
+                                type: integer
+                              thresholdMetPercentage:
+                                description: |-
+                                  ThresholdMetPercentage is the percentage of time the NUMA's system CPU pressure
+                                  must be above the threshold for an eviction to be triggered.
+                                type: number
+                            type: object
                         type: object
                       cpuSystemPressureEvictionConfig:
                         description: CPUSystemPressureEvictionConfig is the config
@@ -384,6 +436,14 @@ spec:
                             description: EnableSystemLevelEviction is whether to enable
                               system-level eviction
                             type: boolean
+                          evictNonReclaimedAnnotationSelector:
+                            description: EvictNonReclaimedAnnotationSelector is a
+                              non-reclaimed pod eviction anno selector
+                            type: string
+                          evictNonReclaimedLabelSelector:
+                            description: EvictNonReclaimedLabelSelector is a non-reclaimed
+                              pod eviction label selector
+                            type: string
                           gracePeriod:
                             description: GracePeriod is the grace period of memory
                               pressure eviction
@@ -405,12 +465,24 @@ spec:
                               type: string
                             minItems: 1
                             type: array
+                          numaFreeBelowWatermarkTimesReclaimedThreshold:
+                            description: |-
+                              NumaFreeBelowWatermarkTimesReclaimedThreshold is the threshold for the number of
+                              times NUMA's free memory of the reclaimed instance falls below the watermark
+                            minimum: 0
+                            type: integer
                           numaFreeBelowWatermarkTimesThreshold:
                             description: |-
                               NumaFreeBelowWatermarkTimesThreshold is the threshold for the number of
                               times NUMA's free memory falls below the watermark
                             minimum: 0
                             type: integer
+                          numaFreeConstraintFastEvictionWaitCycle:
+                            description: NumaFreeConstraintFastEvictionWaitCycle is
+                              the wait cycle for fast eviction when numa memory is
+                              extremely tight
+                            minimum: 0
+                            type: integer
                           numaVictimMinimumUtilizationThreshold:
                             description: |-
                               NumaVictimMinimumUtilizationThreshold is the victim's minimum memory usage on a NUMA node, if a pod
@@ -641,6 +713,42 @@ spec:
                             type: number
                         type: object
                     type: object
+                  fineGrainedResourceConfig:
+                    description: FineGrainedResourceConfig is a configuration for
+                      more fine-grained resource control
+                    properties:
+                      cpuBurstConfig:
+                        description: CPUBurstConfig has cpu burst related configurations
+                        properties:
+                          defaultCPUBurstPercent:
+                            description: DefaultCPUBurstPercent is the default cpu
+                              burst percent to be set for pods with dedicated cores.
+                            format: int64
+                            maximum: 100
+                            minimum: 0
+                            type: integer
+                          enableDedicatedCoresDefaultCPUBurst:
+                            description: |-
+                              EnableDedicatedCoresDefaultCPUBurst indicates whether cpu burst should be enabled by default for pods with dedicated cores.
+                              If set to true, it means that cpu burst should be enabled by default for pods with dedicated cores (cpu burst value is calculated and set).
+                              If set to false, it means that cpu burst should be disabled for pods with dedicated cores (cpu burst value is set to 0).
+                              If set to nil, it means that no operation is done on the cpu burst value.
+                            type: boolean
+                        type: object
+                    type: object
+                  qrmPluginConfig:
+                    description: QRMPluginConfig is a configuration for qrm plugin
+                    properties:
+                      cpuPluginConfig:
+                        description: CPUPluginConfig is the config for cpu plugin
+                        properties:
+                          preferUseExistNUMAHintResult:
+                            description: |-
+                              PreferUseExistNUMAHintResult prefer to use existing numa hint results
+                              The calculation results may originate from upstream components and be recorded in the pod annotation
+                            type: boolean
+                        type: object
+                    type: object
                   reclaimedResourceConfig:
                     description: ReclaimedResourceConfig is a configuration for reclaim
                       resource
@@ -676,6 +784,18 @@ spec:
                                 maximum: 1
                                 minimum: 0
                                 type: number
+                              nonReclaimUtilizationHigh:
+                                description: NonReclaimUtilizationHigh is the high
+                                  CPU utilization threshold
+                                maximum: 1
+                                minimum: 0
+                                type: number
+                              nonReclaimUtilizationLow:
+                                description: NonReclaimUtilizationLow is the low CPU
+                                  utilization threshold
+                                maximum: 1
+                                minimum: 0
+                                type: number
                               targetReclaimedCoreUtilization:
                                 description: |-
                                   TargetReclaimedCoreUtilization is the target reclaimed core utilization to be used for
@@ -776,6 +896,30 @@ spec:
                           MinReclaimedResourceForReport.
                           For example, {"cpu": 4, "memory": 5Gi}.
                         type: object
+                      numaMinReclaimedResourceForAllocate:
+                        additionalProperties:
+                          anyOf:
+                          - type: integer
+                          - type: string
+                          pattern: ^(\+|-)?(([0-9]+(\.[0-9]*)?)|(\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\+|-)?(([0-9]+(\.[0-9]*)?)|(\.[0-9]+))))?$
+                          x-kubernetes-int-or-string: true
+                        description: |-
+                          NumaMinReclaimedResourceForAllocate is a resource reserved at numa level for reclaimed_cores pods，these resources will not be used
+                          by shared_cores pods.
+                          For example, {"cpu": 2, "memory": 0Gi}.
+                        type: object
+                      numaMinReclaimedResourceRatioForAllocate:
+                        additionalProperties:
+                          anyOf:
+                          - type: integer
+                          - type: string
+                          pattern: ^(\+|-)?(([0-9]+(\.[0-9]*)?)|(\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\+|-)?(([0-9]+(\.[0-9]*)?)|(\.[0-9]+))))?$
+                          x-kubernetes-int-or-string: true
+                        description: |-
+                          NumaMinReclaimedResourceRatioForAllocate is a resource reserved ratio at numa level for reclaimed_cores pods，these resources will not be used
+                          by shared_cores pods.
+                          For example, {"cpu": 0.1, "memory": 0.2}.
+                        type: object
                       reservedResourceForAllocate:
                         additionalProperties:
                           anyOf:

config/crd/bases/config.katalyst.kubewharf.io_irqtuningconfigurations.yaml

@@ -201,6 +201,9 @@ spec:
                     maximum: 100
                     minimum: 0
                     type: integer
+                  disableXPS:
+                    description: DisableXPS indicates whether to disable the XPS function.
+                    type: boolean
                   enableRPS:
                     description: |-
                       EnableRPS indicates whether to enable the RPS function.
@@ -266,6 +269,12 @@ spec:
                       NICAffinityPolicy represents the NICs's irqs affinity sockets policy.
                       One of CompleteMap, OverallBalance, PhysicalTopo.
                     type: string
+                  normalThroughputNics:
+                    description: NormalThroughputNics describes static configured
+                      normal throughput Nics.
+                    items:
+                      type: string
+                    type: array
                   reniceKsoftirqd:
                     description: ReniceKsoftirqd indicates whether to renice ksoftirqd
                       process.

config/crd/bases/config.katalyst.kubewharf.io_strategygroups.yaml

@@ -228,8 +228,6 @@ spec:
                   this config.
                 format: int32
                 type: integer
-            required:
-            - enabledStrategies
             type: object
         type: object
     served: true