client

func (*Adapter) Create ¶

func (adapter *Adapter) Create(ctx context.Context, r resource.Resource, opt ...state.CreateOption) error

Create a resource.

If a resource already exists, Create returns an error.

func (*Adapter) Destroy ¶

func (adapter *Adapter) Destroy(ctx context.Context, resourcePointer resource.Pointer, opt ...state.DestroyOption) error

Destroy a resource.

If a resource doesn't exist, error is returned. If a resource has pending finalizers, error is returned.

func (*Adapter) Get ¶

func (adapter *Adapter) Get(ctx context.Context, resourcePointer resource.Pointer, opt ...state.GetOption) (resource.Resource, error)

Get a resource by type and ID.

If a resource is not found, error is returned.

func (*Adapter) List ¶

func (adapter *Adapter) List(ctx context.Context, resourceKind resource.Kind, opt ...state.ListOption) (resource.List, error)

List resources by type.

func (*Adapter) Teardown ¶ added in v1.16.0

func (adapter *Adapter) Teardown(ctx context.Context, resourcePointer resource.Pointer, opt ...state.TeardownOption) (bool, error)

Teardown a resource (mark as being destroyed) using a single Teardown RPC.

Implements state.Teardowner, so state.WrapCore over this Adapter routes state.State.Teardown through this single round-trip rather than the default Get + Update fallback.

If the server does not implement the Teardown RPC (returns codes.Unimplemented), this method transparently falls back to the default Get + Update path; the fallback is sticky so subsequent calls skip the round-trip.

func (*Adapter) TeardownAndDestroy ¶ added in v1.16.0

func (adapter *Adapter) TeardownAndDestroy(ctx context.Context, resourcePointer resource.Pointer, opt ...state.TeardownAndDestroyOption) error

TeardownAndDestroy a resource via a single blocking RPC.

Implements state.TeardownAndDestroyer, so state.WrapCore over this Adapter routes state.State.TeardownAndDestroy through one server-side call that handles teardown, the wait for finalizers, and destroy in-process against the wrapped state.

If the server does not implement the TeardownAndDestroy RPC (returns codes.Unimplemented), this method transparently falls back to the default Teardown + WatchFor + Destroy path; the fallback is sticky so subsequent calls skip the round-trip.

func (*Adapter) Update ¶

func (adapter *Adapter) Update(ctx context.Context, newResource resource.Resource, opt ...state.UpdateOption) error

Update a resource.

If a resource doesn't exist, error is returned. On update current version of resource `new` in the state should match the version on the backend, otherwise conflict error is returned.

func (*Adapter) Watch ¶

func (adapter *Adapter) Watch(ctx context.Context, resourcePointer resource.Pointer, ch chan<- state.Event, opt ...state.WatchOption) error

Watch state of a resource by type.

It's fine to watch for a resource which doesn't exist yet. Watch is canceled when context gets canceled. Watch sends initial resource state as the very first event on the channel, and then sends any updates to the resource as events.

func (*Adapter) WatchKind ¶

func (adapter *Adapter) WatchKind(ctx context.Context, resourceKind resource.Kind, ch chan<- state.Event, opt ...state.WatchKindOption) error

WatchKind watches resources of specific kind (namespace and type).

func (*Adapter) WatchKindAggregated ¶ added in v0.3.0

func (adapter *Adapter) WatchKindAggregated(ctx context.Context, resourceKind resource.Kind, ch chan<- []state.Event, opt ...state.WatchKindOption) error

WatchKindAggregated watches resources of specific kind (namespace and type).