runtime_details

Documentation

Index

• func GetPodGeneration(ctx context.Context, kubeClient *kubernetes.Clientset, pod *corev1.Pod) (int64, error)
• func GetRuntimeDetails(ctx context.Context, kubeClient client.Client, ...) (*odigosv1.InstrumentedApplication, error)
• func SetupWithManager(mgr ctrl.Manager, clientset *kubernetes.Clientset) error
• type DeprecatedInstrumentationConfigReconciler
  • func (i *DeprecatedInstrumentationConfigReconciler) Reconcile(ctx context.Context, request reconcile.Request) (reconcile.Result, error)
• type InstrumentationConfigReconciler
  • func (r *InstrumentationConfigReconciler) Reconcile(ctx context.Context, request reconcile.Request) (reconcile.Result, error)
• type PodsReconciler
  • func (p *PodsReconciler) Reconcile(ctx context.Context, request reconcile.Request) (reconcile.Result, error)

Functions

func GetPodGeneration

func GetPodGeneration(ctx context.Context, kubeClient *kubernetes.Clientset, pod *corev1.Pod) (int64, error)

Givin a pod, this function calculates it's workload generation. e.g. it will return the generation of the deployment/ds/ss manifest that created the pod. This function uses clientset and not the controller-runtime client, so not to pull these objects into the cache, potentially increasing the memory usage of the process. Avoid calling this function too often, as it will make an API call to the k8s API server.

func GetRuntimeDetails

func GetRuntimeDetails(ctx context.Context, kubeClient client.Client, podWorkload *workload.PodWorkload) (*odigosv1.InstrumentedApplication, error)

func SetupWithManager

func SetupWithManager(mgr ctrl.Manager, clientset *kubernetes.Clientset) error

Types

type DeprecatedInstrumentationConfigReconciler

type DeprecatedInstrumentationConfigReconciler struct {
	client.Client
	Scheme *runtime.Scheme
}

replaced by odiglet/pkg/kube/runtime_details/instrumentationconfigs_controller.go which does not rely on the RuntimeDetailsInvalidated flag left here until the migration is complete Deprecated: the new runtime inspection logic is found in odiglet/pkg/kube/runtime_details/instrumentationconfigs_controller.go

func (*DeprecatedInstrumentationConfigReconciler) Reconcile

func (i *DeprecatedInstrumentationConfigReconciler) Reconcile(ctx context.Context, request reconcile.Request) (reconcile.Result, error)

type InstrumentationConfigReconciler

type InstrumentationConfigReconciler struct {
	client.Client
	Scheme *runtime.Scheme

	// the clientset is used to interact with the k8s API directly,
	// without pulling in specific objects into the controller runtime cache
	// which can be expensive (memory and CPU)
	Clientset *kubernetes.Clientset
}

func (*InstrumentationConfigReconciler) Reconcile

func (r *InstrumentationConfigReconciler) Reconcile(ctx context.Context, request reconcile.Request) (reconcile.Result, error)

type PodsReconciler

type PodsReconciler struct {
	client.Client
	Scheme *runtime.Scheme

	// the clientset is used to interact with the k8s API directly,
	// without pulling in specific objects into the controller runtime cache
	// which can be expensive (memory and CPU)
	Clientset *kubernetes.Clientset
}

func (*PodsReconciler) Reconcile

func (p *PodsReconciler) Reconcile(ctx context.Context, request reconcile.Request) (reconcile.Result, error)

We need to apply runtime details detection for a new running pod in the following cases: 1. When a new workload generation is applied, the runtime details might be changed (different env, versions, etc). 2. When a source is added, but there are no running pods yet. When the first pod starts running, this is chance to apply runtime details detection.