Add basic skeletons of state handling, containers and pods

Signed-off-by: Matthew Heon <mheon@redhat.com>
This commit is contained in:
Matthew Heon 2017-09-06 11:15:11 -04:00
parent 2850fb60a5
commit 35e951fc8c
5 changed files with 376 additions and 27 deletions

View file

@ -20,6 +20,7 @@ type RuntimeOption func(*Runtime) error
// Runtime is the core libpod runtime
type Runtime struct {
config *RuntimeConfig
state State
store storage.Store
imageContext *types.SystemContext
apparmorEnabled bool
@ -145,45 +146,158 @@ type ContainerFilter func(*Container) bool
// NewContainer creates a new container from a given OCI config
func (r *Runtime) NewContainer(spec *spec.Spec, options ...CtrCreateOption) (*Container, error) {
return nil, ctr.ErrNotImplemented
r.lock.Lock()
defer r.lock.Unlock()
if !r.valid {
return nil, fmt.Errorf("runtime has already been shut down")
}
ctr, err := newContainer(spec)
if err != nil {
return nil, err
}
for _, option := range options {
if err := option(ctr); err != nil {
return nil, errors.Wrapf(err, "error running container create option")
}
}
ctr.valid = true
if err := r.state.AddContainer(ctr); err != nil {
// If we joined a pod, remove ourself from it
if ctr.pod != nil {
if err2 := ctr.pod.removeContainer(ctr); err2 != nil {
return nil, errors.Wrapf(err, "error adding new container to state, container could not be removed from pod %s", ctr.pod.ID())
}
}
// TODO: Might be worth making an effort to detect duplicate IDs
// We can recover from that by generating a new ID for the
// container
return nil, errors.Wrapf(err, "error adding new container to state")
}
return ctr, nil
}
// RemoveContainer removes the given container
// If force is specified, the container will be stopped first
// Otherwise, RemoveContainer will return an error if the container is running
func (r *Runtime) RemoveContainer(c *Container, force bool) error {
return ctr.ErrNotImplemented
return errNotImplemented
}
// GetContainer retrieves a container by its ID
func (r *Runtime) GetContainer(id string) (*ctr.Container, error) {
return nil, ctr.ErrNotImplemented
func (r *Runtime) GetContainer(id string) (*Container, error) {
r.lock.RLock()
defer r.lock.RUnlock()
if !r.valid {
return nil, fmt.Errorf("runtime has already been shut down")
}
return r.state.GetContainer(id)
}
// HasContainer checks if a container with the given ID is present
func (r *Runtime) HasContainer(id string) (bool, error) {
r.lock.RLock()
defer r.lock.RUnlock()
if !r.valid {
return false, fmt.Errorf("runtime has already been shut down")
}
return r.state.HasContainer(id)
}
// LookupContainer looks up a container by its name or a partial ID
// If a partial ID is not unique, an error will be returned
func (r *Runtime) LookupContainer(idOrName string) (*ctr.Container, error) {
return nil, ctr.ErrNotImplemented
func (r *Runtime) LookupContainer(idOrName string) (*Container, error) {
r.lock.RLock()
defer r.lock.RUnlock()
if !r.valid {
return nil, fmt.Errorf("runtime has already been shut down")
}
return r.state.LookupContainer(idOrName)
}
// GetContainers retrieves all containers from the state
// Filters can be provided which will determine what containers are included in
// the output. Multiple filters are handled by ANDing their output, so only
// containers matching all filters are returned
func (r *Runtime) GetContainers(filters ...ContainerFilter) ([]*ctr.Container, error) {
return nil, ctr.ErrNotImplemented
func (r *Runtime) GetContainers(filters ...ContainerFilter) ([]*Container, error) {
r.lock.RLock()
defer r.lock.RUnlock()
if !r.valid {
return nil, fmt.Errorf("runtime has already been shut down")
}
ctrs, err := r.state.GetAllContainers()
if err != nil {
return nil, err
}
ctrsFiltered := make([]*Container, 0, len(ctrs))
for _, ctr := range ctrs {
include := true
for _, filter := range filters {
include = include && filter(ctr)
}
if include {
ctrsFiltered = append(ctrsFiltered, ctr)
}
}
return ctrsFiltered, nil
}
// Pod API
// A PodCreateOption is a functional option which alters the Pod created by
// NewPod
type PodCreateOption func(*Pod) error
// PodFilter is a function to determine whether a pod is included in command
// output. Pods to be outputted are tested using the function. A true return
// will include the pod, a false return will exclude it.
type PodFilter func(*Pod) bool
// NewPod makes a new, empty pod
func (r *Runtime) NewPod() (*pod.Pod, error) {
return nil, ctr.ErrNotImplemented
func (r *Runtime) NewPod(options ...PodCreateOption) (*Pod, error) {
r.lock.Lock()
defer r.lock.Unlock()
if !r.valid {
return nil, fmt.Errorf("runtime has already been shut down")
}
pod, err := newPod()
if err != nil {
return nil, errors.Wrapf(err, "error creating pod")
}
for _, option := range options {
if err := option(pod); err != nil {
return nil, errors.Wrapf(err, "error running pod create option")
}
}
pod.valid = true
if err := r.state.AddPod(pod); err != nil {
return nil, errors.Wrapf(err, "error adding pod to state")
}
return nil, errNotImplemented
}
// RemovePod removes a pod and all containers in it
@ -191,24 +305,74 @@ func (r *Runtime) NewPod() (*pod.Pod, error) {
// Otherwise, RemovePod will return an error if any container in the pod is running
// Remove acts atomically, removing all containers or no containers
func (r *Runtime) RemovePod(p *Pod, force bool) error {
return ctr.ErrNotImplemented
return errNotImplemented
}
// GetPod retrieves a pod by its ID
func (r *Runtime) GetPod(id string) (*pod.Pod, error) {
return nil, ctr.ErrNotImplemented
func (r *Runtime) GetPod(id string) (*Pod, error) {
r.lock.RLock()
defer r.lock.RUnlock()
if !r.valid {
return nil, fmt.Errorf("runtime has already been shut down")
}
return r.state.GetPod(id)
}
// HasPod checks to see if a pod with the given ID exists
func (r *Runtime) HasPod(id string) (bool, error) {
r.lock.RLock()
defer r.lock.RUnlock()
if !r.valid {
return false, fmt.Errorf("runtime has already been shut down")
}
return r.state.HasPod(id)
}
// LookupPod retrieves a pod by its name or a partial ID
// If a partial ID is not unique, an error will be returned
func (r *Runtime) LookupPod(idOrName string) (*pod.Pod, error) {
return nil, ctr.ErrNotImplemented
func (r *Runtime) LookupPod(idOrName string) (*Pod, error) {
r.lock.RLock()
defer r.lock.RUnlock()
if !r.valid {
return nil, fmt.Errorf("runtime has already been shut down")
}
return r.state.LookupPod(idOrName)
}
// GetPods retrieves all pods
// Filters can be provided which will determine which pods are included in the
// output. Multiple filters are handled by ANDing their output, so only pods
// matching all filters are returned
func (r *Runtime) GetPods(filters ...PodFilter) ([]*pod.Pod, error) {
return nil, ctr.ErrNotImplemented
func (r *Runtime) GetPods(filters ...PodFilter) ([]*Pod, error) {
r.lock.RLock()
defer r.lock.RUnlock()
if !r.valid {
return nil, fmt.Errorf("runtime has already been shut down")
}
pods, err := r.state.GetAllPods()
if err != nil {
return nil, err
}
podsFiltered := make([]*Pod, 0, len(pods))
for _, pod := range pods {
include := true
for _, filter := range filters {
include = include && filter(pod)
}
if include {
podsFiltered = append(podsFiltered, pod)
}
}
return podsFiltered, nil
}