abra/pkg/upstream/stack/remove.go

package stack // https://github.com/docker/cli/blob/master/cli/command/stack/remove.go

import (
	"context"
	"fmt"
	"os"
	"os/signal"
	"sort"
	"strings"
	"time"

	"coopcloud.tech/abra/pkg/i18n"
	"coopcloud.tech/abra/pkg/log"
	"github.com/docker/docker/api/types"
	containerPkg "github.com/docker/docker/api/types/container"
	"github.com/docker/docker/api/types/network"
	"github.com/docker/docker/api/types/swarm"
	"github.com/docker/docker/api/types/versions"
	"github.com/docker/docker/client"
	apiclient "github.com/docker/docker/client"
	"github.com/pkg/errors"
)

// RunRemove is the swarm implementation of docker stack remove
func RunRemove(ctx context.Context, client *apiclient.Client, opts Remove) error {
	log.Info(i18n.G("initialising undeploy"))

	if WaitTimeout != 0 {
		log.Debug(i18n.G("timeout: set to %d second(s)", WaitTimeout))
	}

	sigIntCh := make(chan os.Signal, 1)
	signal.Notify(sigIntCh, os.Interrupt)
	defer signal.Stop(sigIntCh)

	waitCh := make(chan struct{})
	errCh := make(chan error)

	go func() {
		var errs []string
		for _, namespace := range opts.Namespaces {
			services, err := GetStackServices(ctx, client, namespace)
			if err != nil {
				errCh <- err
				return
			}

			networks, err := getStackNetworks(ctx, client, namespace)
			if err != nil {
				errCh <- err
				return
			}

			var secrets []swarm.Secret
			if versions.GreaterThanOrEqualTo(client.ClientVersion(), "1.25") {
				secrets, err = getStackSecrets(ctx, client, namespace)
				if err != nil {
					errCh <- err
					return
				}
			}

			var configs []swarm.Config
			if versions.GreaterThanOrEqualTo(client.ClientVersion(), "1.30") {
				configs, err = getStackConfigs(ctx, client, namespace)
				if err != nil {
					errCh <- err
					return
				}
			}

			if len(services)+len(networks)+len(secrets)+len(configs) == 0 {
				log.Warn(i18n.G("nothing found in stack: %s", namespace))
				continue
			}

			hasError := removeServices(ctx, client, services)
			hasError = removeSecrets(ctx, client, secrets) || hasError
			hasError = removeConfigs(ctx, client, configs) || hasError
			hasError = removeNetworks(ctx, client, networks) || hasError

			if hasError {
				errs = append(errs, i18n.G("failed to remove some resources from stack: %s", namespace))
				continue
			}

			log.Info(i18n.G("polling undeploy status"))
			timeout, err := waitOnTasks(ctx, client, namespace)
			if timeout {
				errs = append(errs, err.Error())
			} else {
				if err != nil {
					errs = append(errs, fmt.Sprint(i18n.G("failed to wait on tasks of stack: %s: %s", namespace, err)))
				}
			}
		}

		if len(errs) > 0 {
			errCh <- errors.New(strings.Join(errs, "\n"))
			return
		}

		close(waitCh)
	}()

	select {
	case <-waitCh:
		return nil
	case <-sigIntCh:
		return errors.New(i18n.G("skipping as requested, undeploy still in progress 🟠"))
	case err := <-errCh:
		return err
	}

	return nil
}

func sortServiceByName(services []swarm.Service) func(i, j int) bool {
	return func(i, j int) bool {
		return services[i].Spec.Name < services[j].Spec.Name
	}
}

func removeServices(
	ctx context.Context,
	client *apiclient.Client,
	services []swarm.Service,
) bool {
	var hasError bool
	sort.Slice(services, sortServiceByName(services))
	for _, service := range services {
		log.Debug(i18n.G("removing service %s", service.Spec.Name))
		if err := client.ServiceRemove(ctx, service.ID); err != nil {
			hasError = true
			log.Fatal(i18n.G("failed to remove service %s: %s", service.ID, err))
		}
	}
	return hasError
}

func removeNetworks(
	ctx context.Context,
	client *apiclient.Client,
	networks []network.Inspect,
) bool {
	var hasError bool
	for _, network := range networks {
		log.Debug(i18n.G("removing network %s", network.Name))
		if err := client.NetworkRemove(ctx, network.ID); err != nil {
			hasError = true
			log.Fatal(i18n.G("failed to remove network %s: %s", network.ID, err))
		}
	}
	return hasError
}

func removeSecrets(
	ctx context.Context,
	client *apiclient.Client,
	secrets []swarm.Secret,
) bool {
	var hasError bool
	for _, secret := range secrets {
		log.Debug(i18n.G("removing secret %s", secret.Spec.Name))
		if err := client.SecretRemove(ctx, secret.ID); err != nil {
			hasError = true
			log.Fatal(i18n.G("failed to remove secret %s: %s", secret.ID, err))
		}
	}
	return hasError
}

func removeConfigs(
	ctx context.Context,
	client *apiclient.Client,
	configs []swarm.Config,
) bool {
	var hasError bool
	for _, config := range configs {
		log.Debug(i18n.G("removing config %s", config.Spec.Name))
		if err := client.ConfigRemove(ctx, config.ID); err != nil {
			hasError = true
			log.Fatal(i18n.G("failed to remove config %s: %s", config.ID, err))
		}
	}
	return hasError
}

// https://github.com/docker/cli/pull/4259
func getStackTasks(ctx context.Context, apiclient client.APIClient, namespace string) ([]swarm.Task, error) {
	return apiclient.TaskList(ctx, types.TaskListOptions{Filters: getStackFilter(namespace)})
}

func getStackContainers(ctx context.Context, apiclient client.APIClient, namespace string) ([]containerPkg.Summary, error) {
	return apiclient.ContainerList(ctx, containerPkg.ListOptions{Filters: getStackFilter(namespace)})
}

var numberedStates = map[swarm.TaskState]int64{
	swarm.TaskStateNew:       1,
	swarm.TaskStateAllocated: 2,
	swarm.TaskStatePending:   3,
	swarm.TaskStateAssigned:  4,
	swarm.TaskStateAccepted:  5,
	swarm.TaskStatePreparing: 6,
	swarm.TaskStateReady:     7,
	swarm.TaskStateStarting:  8,
	swarm.TaskStateRunning:   9,
	swarm.TaskStateComplete:  10,
	swarm.TaskStateShutdown:  11,
	swarm.TaskStateFailed:    12,
	swarm.TaskStateRejected:  13,
}

func terminalState(state swarm.TaskState) bool {
	return numberedStates[state] > numberedStates[swarm.TaskStateRunning]
}

func waitOnTasks(ctx context.Context, client apiclient.APIClient, namespace string) (bool, error) {
	var timedOut bool

	go func() {
		if WaitTimeout == 0 {
			return
		}

		log.Debug(i18n.G("timeout: waiting on undeploy tasks (timeout=%v secs)", WaitTimeout))

		timeout := time.Duration(WaitTimeout) * time.Second
		<-time.After(timeout)

		log.Debug(i18n.G("timed out on undeploy (timeout=%v sec)", WaitTimeout))
		timedOut = true
	}()

	terminalStatesReached := 0
	for {
		tasks, err := getStackTasks(ctx, client, namespace)
		if err != nil {
			return false, errors.New(i18n.G("failed to get tasks: %w", err))
		}

		for _, task := range tasks {
			if terminalState(task.Status.State) {
				log.Debug(i18n.G("waiting for %d task(s) to reach terminal state", len(tasks)-terminalStatesReached))
				terminalStatesReached++
				break
			}
		}

		if terminalStatesReached == len(tasks) {
			log.Debug(i18n.G("all tasks reached terminal state"))
			break
		}

		if timedOut {
			return true, errors.New(i18n.G("deployment timed out 🟠"))
		}
	}

	lastSeenCount := -1
	for {
		containers, err := getStackContainers(ctx, client, namespace)
		if err != nil {
			return false, errors.New(i18n.G("failed to list containers of stack: %s", namespace))
		}

		numContainers := len(containers)

		if numContainers == 0 {
			log.Debug(i18n.G("all containers did really go away"))
			break
		}

		if numContainers != lastSeenCount {
			log.Debug(i18n.G("waiting for %d container(s) to really go away", numContainers))
			lastSeenCount = numContainers
		}
	}

	return false, nil
}