podman/libpod/pod.go

package libpod

import (
	"context"
	"path/filepath"
	"strings"

	"github.com/containers/storage"
	"github.com/docker/docker/pkg/stringid"
	"github.com/pkg/errors"
	"github.com/sirupsen/logrus"
)

// Pod represents a group of containers that may share namespaces
// ffjson: skip
type Pod struct {
	config *PodConfig
	state  *podState

	valid   bool
	runtime *Runtime
	lock    storage.Locker
}

// PodConfig represents a pod's static configuration
type PodConfig struct {
	ID   string `json:"id"`
	Name string `json:"name"`

	// Labels contains labels applied to the pod
	Labels map[string]string `json:"labels"`
	// CgroupParent contains the pod's CGroup parent
	CgroupParent string `json:"cgroupParent"`
	// UsePodCgroup indicates whether the pod will create its own CGroup and
	// join containers to it.
	// If true, all containers joined to the pod will use the pod cgroup as
	// their cgroup parent, and cannot set a different cgroup parent
	UsePodCgroup bool
}

// podState represents a pod's state
type podState struct {
	// CgroupPath is the path to the pod's CGroup
	CgroupPath string
}

// ID retrieves the pod's ID
func (p *Pod) ID() string {
	return p.config.ID
}

// Name retrieves the pod's name
func (p *Pod) Name() string {
	return p.config.Name
}

// Labels returns the pod's labels
func (p *Pod) Labels() map[string]string {
	labels := make(map[string]string)
	for key, value := range p.config.Labels {
		labels[key] = value
	}

	return labels
}

// CgroupParent returns the pod's CGroup parent
func (p *Pod) CgroupParent() string {
	return p.config.CgroupParent
}

// UsePodCgroup returns whether containers in the pod will default to this pod's
// cgroup instead of the default libpod parent
func (p *Pod) UsePodCgroup() bool {
	return p.config.UsePodCgroup
}

// CgroupPath returns the path to the pod's CGroup
func (p *Pod) CgroupPath() (string, error) {
	p.lock.Lock()
	p.lock.Unlock()
	if err := p.updatePod(); err != nil {
		return "", err
	}

	return p.state.CgroupPath, nil
}

// Creates a new, empty pod
func newPod(lockDir string, runtime *Runtime) (*Pod, error) {
	pod := new(Pod)
	pod.config = new(PodConfig)
	pod.config.ID = stringid.GenerateNonCryptoID()
	pod.config.Labels = make(map[string]string)
	pod.state = new(podState)
	pod.runtime = runtime

	// Path our lock file will reside at
	lockPath := filepath.Join(lockDir, pod.config.ID)
	// Grab a lockfile at the given path
	lock, err := storage.GetLockfile(lockPath)
	if err != nil {
		return nil, errors.Wrapf(err, "error creating lockfile for new pod")
	}
	pod.lock = lock

	return pod, nil
}

// Update pod state from database
func (p *Pod) updatePod() error {
	if err := p.runtime.state.UpdatePod(p); err != nil {
		return err
	}

	return nil
}

// Save pod state to database
func (p *Pod) save() error {
	if err := p.runtime.state.SavePod(p); err != nil {
		return errors.Wrapf(err, "error saving pod %s state")
	}

	return nil
}

// Refresh a pod's state after restart
func (p *Pod) refresh() error {
	p.lock.Lock()
	defer p.lock.Unlock()

	if !p.valid {
		return ErrPodRemoved
	}

	// We need to recreate the pod's cgroup
	if p.config.UsePodCgroup {
		switch p.runtime.config.CgroupManager {
		case SystemdCgroupsManager:
			// NOOP for now, until proper systemd cgroup management
			// is implemented
		case CgroupfsCgroupsManager:
			p.state.CgroupPath = filepath.Join(p.config.CgroupParent, p.ID())

			logrus.Debugf("setting pod cgroup to %s", p.state.CgroupPath)
		default:
			return errors.Wrapf(ErrInvalidArg, "unknown cgroups manager %s specified", p.runtime.config.CgroupManager)
		}
	}

	// Save changes
	return p.save()
}

// Start starts all containers within a pod
// It combines the effects of Init() and Start() on a container
// If a container has already been initialized it will be started,
// otherwise it will be initialized then started.
// Containers that are already running or have been paused are ignored
// All containers are started independently, in order dictated by their
// dependencies.
// An error and a map[string]error are returned
// If the error is not nil and the map is nil, an error was encountered before
// any containers were started
// If map is not nil, an error was encountered when starting one or more
// containers. The container ID is mapped to the error encountered. The error is
// set to ErrCtrExists
// If both error and the map are nil, all containers were started successfully
func (p *Pod) Start(ctx context.Context) (map[string]error, error) {
	p.lock.Lock()
	defer p.lock.Unlock()

	if !p.valid {
		return nil, ErrPodRemoved
	}

	allCtrs, err := p.runtime.state.PodContainers(p)
	if err != nil {
		return nil, err
	}

	// Build a dependency graph of containers in the pod
	graph, err := buildContainerGraph(allCtrs)
	if err != nil {
		return nil, errors.Wrapf(err, "error generating dependency graph for pod %s", p.ID())
	}

	ctrErrors := make(map[string]error)
	ctrsVisited := make(map[string]bool)

	// If there are no containers without dependencies, we can't start
	// Error out
	if len(graph.noDepNodes) == 0 {
		return nil, errors.Wrapf(ErrNoSuchCtr, "no containers in pod %s have no dependencies, cannot start pod", p.ID())
	}

	// Traverse the graph beginning at nodes with no dependencies
	for _, node := range graph.noDepNodes {
		startNode(ctx, node, false, ctrErrors, ctrsVisited)
	}

	return ctrErrors, nil
}

// Visit a node on a container graph and start the container, or set an error if
// a dependency failed to start
func startNode(ctx context.Context, node *containerNode, setError bool, ctrErrors map[string]error, ctrsVisited map[string]bool) {
	// First, check if we have already visited the node
	if ctrsVisited[node.id] {
		return
	}

	// If setError is true, a dependency of us failed
	// Mark us as failed and recurse
	if setError {
		// Mark us as visited, and set an error
		ctrsVisited[node.id] = true
		ctrErrors[node.id] = errors.Wrapf(ErrCtrStateInvalid, "a dependency of container %s failed to start", node.id)

		// Hit anyone who depends on us, and set errors on them too
		for _, successor := range node.dependedOn {
			startNode(ctx, successor, true, ctrErrors, ctrsVisited)
		}

		return
	}

	// Have all our dependencies started?
	// If not, don't visit the node yet
	depsVisited := true
	for _, dep := range node.dependsOn {
		depsVisited = depsVisited && ctrsVisited[dep.id]
	}
	if !depsVisited {
		// Don't visit us yet, all dependencies are not up
		// We'll hit the dependencies eventually, and when we do it will
		// recurse here
		return
	}

	// Going to try to start the container, mark us as visited
	ctrsVisited[node.id] = true

	ctrErrored := false

	// Check if dependencies are running
	// Graph traversal means we should have started them
	// But they could have died before we got here
	// Does not require that the container be locked, we only need to lock
	// the dependencies
	depsStopped, err := node.container.checkDependenciesRunning()
	if err != nil {
		ctrErrors[node.id] = err
		ctrErrored = true
	} else if len(depsStopped) > 0 {
		// Our dependencies are not running
		depsList := strings.Join(depsStopped, ",")
		ctrErrors[node.id] = errors.Wrapf(ErrCtrStateInvalid, "the following dependencies of container %s are not running: %s", node.id, depsList)
		ctrErrored = true
	}

	// Lock before we start
	node.container.lock.Lock()

	// Sync the container to pick up current state
	if !ctrErrored {
		if err := node.container.syncContainer(); err != nil {
			ctrErrored = true
			ctrErrors[node.id] = err
		}
	}

	// Start the container (only if it is not running)
	if !ctrErrored && node.container.state.State != ContainerStateRunning {
		if err := node.container.initAndStart(ctx); err != nil {
			ctrErrored = true
			ctrErrors[node.id] = err
		}
	}

	node.container.lock.Unlock()

	// Recurse to anyone who depends on us and start them
	for _, successor := range node.dependedOn {
		startNode(ctx, successor, ctrErrored, ctrErrors, ctrsVisited)
	}

	return
}

// Stop stops all containers within a pod that are not already stopped
// Each container will use its own stop timeout
// Only running containers will be stopped. Paused, stopped, or created
// containers will be ignored.
// If cleanup is true, mounts and network namespaces will be cleaned up after
// the container is stopped.
// All containers are stopped independently. An error stopping one container
// will not prevent other containers being stopped.
// An error and a map[string]error are returned
// If the error is not nil and the map is nil, an error was encountered before
// any containers were stopped
// If map is not nil, an error was encountered when stopping one or more
// containers. The container ID is mapped to the error encountered. The error is
// set to ErrCtrExists
// If both error and the map are nil, all containers were stopped without error
func (p *Pod) Stop(cleanup bool) (map[string]error, error) {
	p.lock.Lock()
	defer p.lock.Unlock()

	if !p.valid {
		return nil, ErrPodRemoved
	}

	allCtrs, err := p.runtime.state.PodContainers(p)
	if err != nil {
		return nil, err
	}

	ctrErrors := make(map[string]error)

	// TODO: There may be cases where it makes sense to order stops based on
	// dependencies. Should we bother with this?

	// Stop to all containers
	for _, ctr := range allCtrs {
		ctr.lock.Lock()

		if err := ctr.syncContainer(); err != nil {
			ctr.lock.Unlock()
			ctrErrors[ctr.ID()] = err
			continue
		}

		// Ignore containers that are not running
		if ctr.state.State != ContainerStateRunning {
			ctr.lock.Unlock()
			continue
		}

		if err := ctr.stop(ctr.config.StopTimeout); err != nil {
			ctr.lock.Unlock()
			ctrErrors[ctr.ID()] = err
			continue
		}

		if cleanup {
			if err := ctr.cleanup(); err != nil {
				ctrErrors[ctr.ID()] = err
			}
		}

		ctr.lock.Unlock()
	}

	if len(ctrErrors) > 0 {
		return ctrErrors, errors.Wrapf(ErrCtrExists, "error stopping some containers")
	}

	return nil, nil
}

// Kill sends a signal to all running containers within a pod
// Signals will only be sent to running containers. Containers that are not
// running will be ignored. All signals are sent independently, and sending will
// continue even if some containers encounter errors.
// An error and a map[string]error are returned
// If the error is not nil and the map is nil, an error was encountered before
// any containers were signalled
// If map is not nil, an error was encountered when signalling one or more
// containers. The container ID is mapped to the error encountered. The error is
// set to ErrCtrExists
// If both error and the map are nil, all containers were signalled successfully
func (p *Pod) Kill(signal uint) (map[string]error, error) {
	p.lock.Lock()
	defer p.lock.Unlock()

	if !p.valid {
		return nil, ErrPodRemoved
	}

	allCtrs, err := p.runtime.state.PodContainers(p)
	if err != nil {
		return nil, err
	}

	ctrErrors := make(map[string]error)

	// Send a signal to all containers
	for _, ctr := range allCtrs {
		ctr.lock.Lock()

		if err := ctr.syncContainer(); err != nil {
			ctr.lock.Unlock()
			ctrErrors[ctr.ID()] = err
			continue
		}

		// Ignore containers that are not running
		if ctr.state.State != ContainerStateRunning {
			ctr.lock.Unlock()
			continue
		}

		if err := ctr.runtime.ociRuntime.killContainer(ctr, signal); err != nil {
			ctr.lock.Unlock()
			ctrErrors[ctr.ID()] = err
			continue
		}

		logrus.Debugf("Killed container %s with signal %d", ctr.ID(), signal)
	}

	if len(ctrErrors) > 0 {
		return ctrErrors, nil
	}

	return nil, nil
}

// HasContainer checks if a container is present in the pod
func (p *Pod) HasContainer(id string) (bool, error) {
	if !p.valid {
		return false, ErrPodRemoved
	}

	return p.runtime.state.PodHasContainer(p, id)
}

// AllContainersByID returns the container IDs of all the containers in the pod
func (p *Pod) AllContainersByID() ([]string, error) {
	p.lock.Lock()
	defer p.lock.Unlock()

	if !p.valid {
		return nil, ErrPodRemoved
	}

	return p.runtime.state.PodContainersByID(p)
}

// AllContainers retrieves the containers in the pod
func (p *Pod) AllContainers() ([]*Container, error) {
	p.lock.Lock()
	defer p.lock.Unlock()

	if !p.valid {
		return nil, ErrPodRemoved
	}

	return p.runtime.state.PodContainers(p)
}

// Status gets the status of all containers in the pod
// Returns a map of Container ID to Container Status
func (p *Pod) Status() (map[string]ContainerStatus, error) {
	p.lock.Lock()
	defer p.lock.Unlock()

	if !p.valid {
		return nil, ErrPodRemoved
	}

	allCtrs, err := p.runtime.state.PodContainers(p)
	if err != nil {
		return nil, err
	}

	// We need to lock all the containers
	for _, ctr := range allCtrs {
		ctr.lock.Lock()
		defer ctr.lock.Unlock()
	}

	// Now that all containers are locked, get their status
	status := make(map[string]ContainerStatus, len(allCtrs))
	for _, ctr := range allCtrs {
		if err := ctr.syncContainer(); err != nil {
			return nil, err
		}

		status[ctr.ID()] = ctr.state.State
	}

	return status, nil
}

// TODO add pod batching
// Lock pod to avoid lock contention
// Store and lock all containers (no RemoveContainer in batch guarantees cache will not become stale)