1
0
Fork 0
mirror of https://github.com/moby/moby.git synced 2022-11-09 12:21:53 -05:00
moby--moby/cli/command/service/update.go
Aaron Lehmann 6d4b527699 Service update failure thresholds and rollback
This adds support for two enhancements to swarm service rolling updates:

- Failure thresholds: In Docker 1.12, a service update could be set up
  to either pause or continue after a single failure occurs. This adds
  an --update-max-failure-ratio flag that controls how many tasks need to
  fail to update for the update as a whole to be considered a failure. A
  counterpart flag, --update-monitor, controls how long to monitor each
  task for a failure after starting it during the update.

- Rollback flag: service update --rollback reverts the service to its
  previous version. If a service update encounters task failures, or
  fails to function properly for some other reason, the user can roll back
  the update.

SwarmKit also has the ability to roll back updates automatically after
hitting the failure thresholds, but we've decided not to expose this in
the Docker API/CLI for now, favoring a workflow where the decision to
roll back is always made by an admin. Depending on user feedback, we may
add a "rollback" option to --update-failure-action in the future.

Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
2016-10-18 10:09:50 -07:00

530 lines
15 KiB
Go

package service
import (
"fmt"
"sort"
"strings"
"time"
"golang.org/x/net/context"
"github.com/docker/docker/api/types"
mounttypes "github.com/docker/docker/api/types/mount"
"github.com/docker/docker/api/types/swarm"
"github.com/docker/docker/cli"
"github.com/docker/docker/cli/command"
"github.com/docker/docker/opts"
runconfigopts "github.com/docker/docker/runconfig/opts"
"github.com/docker/go-connections/nat"
shlex "github.com/flynn-archive/go-shlex"
"github.com/spf13/cobra"
"github.com/spf13/pflag"
)
func newUpdateCommand(dockerCli *command.DockerCli) *cobra.Command {
opts := newServiceOptions()
cmd := &cobra.Command{
Use: "update [OPTIONS] SERVICE",
Short: "Update a service",
Args: cli.ExactArgs(1),
RunE: func(cmd *cobra.Command, args []string) error {
return runUpdate(dockerCli, cmd.Flags(), args[0])
},
}
flags := cmd.Flags()
flags.String("image", "", "Service image tag")
flags.String("args", "", "Service command args")
flags.Bool("rollback", false, "Rollback to previous specification")
addServiceFlags(cmd, opts)
flags.Var(newListOptsVar(), flagEnvRemove, "Remove an environment variable")
flags.Var(newListOptsVar(), flagGroupRemove, "Remove previously added user groups from the container")
flags.Var(newListOptsVar(), flagLabelRemove, "Remove a label by its key")
flags.Var(newListOptsVar(), flagContainerLabelRemove, "Remove a container label by its key")
flags.Var(newListOptsVar(), flagMountRemove, "Remove a mount by its target path")
flags.Var(newListOptsVar(), flagPublishRemove, "Remove a published port by its target port")
flags.Var(newListOptsVar(), flagConstraintRemove, "Remove a constraint")
flags.Var(&opts.labels, flagLabelAdd, "Add or update service labels")
flags.Var(&opts.containerLabels, flagContainerLabelAdd, "Add or update container labels")
flags.Var(&opts.env, flagEnvAdd, "Add or update environment variables")
flags.Var(&opts.mounts, flagMountAdd, "Add or update a mount on a service")
flags.StringSliceVar(&opts.constraints, flagConstraintAdd, []string{}, "Add or update placement constraints")
flags.Var(&opts.endpoint.ports, flagPublishAdd, "Add or update a published port")
return cmd
}
func newListOptsVar() *opts.ListOpts {
return opts.NewListOptsRef(&[]string{}, nil)
}
func runUpdate(dockerCli *command.DockerCli, flags *pflag.FlagSet, serviceID string) error {
apiClient := dockerCli.Client()
ctx := context.Background()
updateOpts := types.ServiceUpdateOptions{}
service, _, err := apiClient.ServiceInspectWithRaw(ctx, serviceID)
if err != nil {
return err
}
rollback, err := flags.GetBool("rollback")
if err != nil {
return err
}
spec := &service.Spec
if rollback {
spec = service.PreviousSpec
if spec == nil {
return fmt.Errorf("service does not have a previous specification to roll back to")
}
}
err = updateService(flags, spec)
if err != nil {
return err
}
// only send auth if flag was set
sendAuth, err := flags.GetBool(flagRegistryAuth)
if err != nil {
return err
}
if sendAuth {
// Retrieve encoded auth token from the image reference
// This would be the old image if it didn't change in this update
image := spec.TaskTemplate.ContainerSpec.Image
encodedAuth, err := command.RetrieveAuthTokenFromImage(ctx, dockerCli, image)
if err != nil {
return err
}
updateOpts.EncodedRegistryAuth = encodedAuth
} else if rollback {
updateOpts.RegistryAuthFrom = types.RegistryAuthFromPreviousSpec
} else {
updateOpts.RegistryAuthFrom = types.RegistryAuthFromSpec
}
err = apiClient.ServiceUpdate(ctx, service.ID, service.Version, *spec, updateOpts)
if err != nil {
return err
}
fmt.Fprintf(dockerCli.Out(), "%s\n", serviceID)
return nil
}
func updateService(flags *pflag.FlagSet, spec *swarm.ServiceSpec) error {
updateString := func(flag string, field *string) {
if flags.Changed(flag) {
*field, _ = flags.GetString(flag)
}
}
updateInt64Value := func(flag string, field *int64) {
if flags.Changed(flag) {
*field = flags.Lookup(flag).Value.(int64Value).Value()
}
}
updateFloat32 := func(flag string, field *float32) {
if flags.Changed(flag) {
*field, _ = flags.GetFloat32(flag)
}
}
updateDuration := func(flag string, field *time.Duration) {
if flags.Changed(flag) {
*field, _ = flags.GetDuration(flag)
}
}
updateDurationOpt := func(flag string, field **time.Duration) {
if flags.Changed(flag) {
val := *flags.Lookup(flag).Value.(*DurationOpt).Value()
*field = &val
}
}
updateUint64 := func(flag string, field *uint64) {
if flags.Changed(flag) {
*field, _ = flags.GetUint64(flag)
}
}
updateUint64Opt := func(flag string, field **uint64) {
if flags.Changed(flag) {
val := *flags.Lookup(flag).Value.(*Uint64Opt).Value()
*field = &val
}
}
cspec := &spec.TaskTemplate.ContainerSpec
task := &spec.TaskTemplate
taskResources := func() *swarm.ResourceRequirements {
if task.Resources == nil {
task.Resources = &swarm.ResourceRequirements{}
}
return task.Resources
}
updateString(flagName, &spec.Name)
updateLabels(flags, &spec.Labels)
updateContainerLabels(flags, &cspec.Labels)
updateString("image", &cspec.Image)
updateStringToSlice(flags, "args", &cspec.Args)
updateEnvironment(flags, &cspec.Env)
updateString(flagWorkdir, &cspec.Dir)
updateString(flagUser, &cspec.User)
updateMounts(flags, &cspec.Mounts)
if flags.Changed(flagLimitCPU) || flags.Changed(flagLimitMemory) {
taskResources().Limits = &swarm.Resources{}
updateInt64Value(flagLimitCPU, &task.Resources.Limits.NanoCPUs)
updateInt64Value(flagLimitMemory, &task.Resources.Limits.MemoryBytes)
}
if flags.Changed(flagReserveCPU) || flags.Changed(flagReserveMemory) {
taskResources().Reservations = &swarm.Resources{}
updateInt64Value(flagReserveCPU, &task.Resources.Reservations.NanoCPUs)
updateInt64Value(flagReserveMemory, &task.Resources.Reservations.MemoryBytes)
}
updateDurationOpt(flagStopGracePeriod, &cspec.StopGracePeriod)
if anyChanged(flags, flagRestartCondition, flagRestartDelay, flagRestartMaxAttempts, flagRestartWindow) {
if task.RestartPolicy == nil {
task.RestartPolicy = &swarm.RestartPolicy{}
}
if flags.Changed(flagRestartCondition) {
value, _ := flags.GetString(flagRestartCondition)
task.RestartPolicy.Condition = swarm.RestartPolicyCondition(value)
}
updateDurationOpt(flagRestartDelay, &task.RestartPolicy.Delay)
updateUint64Opt(flagRestartMaxAttempts, &task.RestartPolicy.MaxAttempts)
updateDurationOpt(flagRestartWindow, &task.RestartPolicy.Window)
}
if anyChanged(flags, flagConstraintAdd, flagConstraintRemove) {
if task.Placement == nil {
task.Placement = &swarm.Placement{}
}
updatePlacement(flags, task.Placement)
}
if err := updateReplicas(flags, &spec.Mode); err != nil {
return err
}
if anyChanged(flags, flagUpdateParallelism, flagUpdateDelay, flagUpdateMonitor, flagUpdateFailureAction, flagUpdateMaxFailureRatio) {
if spec.UpdateConfig == nil {
spec.UpdateConfig = &swarm.UpdateConfig{}
}
updateUint64(flagUpdateParallelism, &spec.UpdateConfig.Parallelism)
updateDuration(flagUpdateDelay, &spec.UpdateConfig.Delay)
updateDuration(flagUpdateMonitor, &spec.UpdateConfig.Monitor)
updateString(flagUpdateFailureAction, &spec.UpdateConfig.FailureAction)
updateFloat32(flagUpdateMaxFailureRatio, &spec.UpdateConfig.MaxFailureRatio)
}
if flags.Changed(flagEndpointMode) {
value, _ := flags.GetString(flagEndpointMode)
if spec.EndpointSpec == nil {
spec.EndpointSpec = &swarm.EndpointSpec{}
}
spec.EndpointSpec.Mode = swarm.ResolutionMode(value)
}
if anyChanged(flags, flagGroupAdd, flagGroupRemove) {
if err := updateGroups(flags, &cspec.Groups); err != nil {
return err
}
}
if anyChanged(flags, flagPublishAdd, flagPublishRemove) {
if spec.EndpointSpec == nil {
spec.EndpointSpec = &swarm.EndpointSpec{}
}
if err := updatePorts(flags, &spec.EndpointSpec.Ports); err != nil {
return err
}
}
if err := updateLogDriver(flags, &spec.TaskTemplate); err != nil {
return err
}
return nil
}
func updateStringToSlice(flags *pflag.FlagSet, flag string, field *[]string) error {
if !flags.Changed(flag) {
return nil
}
value, _ := flags.GetString(flag)
valueSlice, err := shlex.Split(value)
*field = valueSlice
return err
}
func anyChanged(flags *pflag.FlagSet, fields ...string) bool {
for _, flag := range fields {
if flags.Changed(flag) {
return true
}
}
return false
}
func updatePlacement(flags *pflag.FlagSet, placement *swarm.Placement) {
field, _ := flags.GetStringSlice(flagConstraintAdd)
placement.Constraints = append(placement.Constraints, field...)
toRemove := buildToRemoveSet(flags, flagConstraintRemove)
placement.Constraints = removeItems(placement.Constraints, toRemove, itemKey)
}
func updateContainerLabels(flags *pflag.FlagSet, field *map[string]string) {
if flags.Changed(flagContainerLabelAdd) {
if *field == nil {
*field = map[string]string{}
}
values := flags.Lookup(flagContainerLabelAdd).Value.(*opts.ListOpts).GetAll()
for key, value := range runconfigopts.ConvertKVStringsToMap(values) {
(*field)[key] = value
}
}
if *field != nil && flags.Changed(flagContainerLabelRemove) {
toRemove := flags.Lookup(flagContainerLabelRemove).Value.(*opts.ListOpts).GetAll()
for _, label := range toRemove {
delete(*field, label)
}
}
}
func updateLabels(flags *pflag.FlagSet, field *map[string]string) {
if flags.Changed(flagLabelAdd) {
if *field == nil {
*field = map[string]string{}
}
values := flags.Lookup(flagLabelAdd).Value.(*opts.ListOpts).GetAll()
for key, value := range runconfigopts.ConvertKVStringsToMap(values) {
(*field)[key] = value
}
}
if *field != nil && flags.Changed(flagLabelRemove) {
toRemove := flags.Lookup(flagLabelRemove).Value.(*opts.ListOpts).GetAll()
for _, label := range toRemove {
delete(*field, label)
}
}
}
func updateEnvironment(flags *pflag.FlagSet, field *[]string) {
envSet := map[string]string{}
for _, v := range *field {
envSet[envKey(v)] = v
}
if flags.Changed(flagEnvAdd) {
value := flags.Lookup(flagEnvAdd).Value.(*opts.ListOpts)
for _, v := range value.GetAll() {
envSet[envKey(v)] = v
}
}
*field = []string{}
for _, v := range envSet {
*field = append(*field, v)
}
toRemove := buildToRemoveSet(flags, flagEnvRemove)
*field = removeItems(*field, toRemove, envKey)
}
func envKey(value string) string {
kv := strings.SplitN(value, "=", 2)
return kv[0]
}
func itemKey(value string) string {
return value
}
func buildToRemoveSet(flags *pflag.FlagSet, flag string) map[string]struct{} {
var empty struct{}
toRemove := make(map[string]struct{})
if !flags.Changed(flag) {
return toRemove
}
toRemoveSlice := flags.Lookup(flag).Value.(*opts.ListOpts).GetAll()
for _, key := range toRemoveSlice {
toRemove[key] = empty
}
return toRemove
}
func removeItems(
seq []string,
toRemove map[string]struct{},
keyFunc func(string) string,
) []string {
newSeq := []string{}
for _, item := range seq {
if _, exists := toRemove[keyFunc(item)]; !exists {
newSeq = append(newSeq, item)
}
}
return newSeq
}
func updateMounts(flags *pflag.FlagSet, mounts *[]mounttypes.Mount) {
if flags.Changed(flagMountAdd) {
values := flags.Lookup(flagMountAdd).Value.(*MountOpt).Value()
*mounts = append(*mounts, values...)
}
toRemove := buildToRemoveSet(flags, flagMountRemove)
newMounts := []mounttypes.Mount{}
for _, mount := range *mounts {
if _, exists := toRemove[mount.Target]; !exists {
newMounts = append(newMounts, mount)
}
}
*mounts = newMounts
}
func updateGroups(flags *pflag.FlagSet, groups *[]string) error {
if flags.Changed(flagGroupAdd) {
values, err := flags.GetStringSlice(flagGroupAdd)
if err != nil {
return err
}
*groups = append(*groups, values...)
}
toRemove := buildToRemoveSet(flags, flagGroupRemove)
newGroups := []string{}
for _, group := range *groups {
if _, exists := toRemove[group]; !exists {
newGroups = append(newGroups, group)
}
}
// Sort so that result is predictable.
sort.Strings(newGroups)
*groups = newGroups
return nil
}
type byPortConfig []swarm.PortConfig
func (r byPortConfig) Len() int { return len(r) }
func (r byPortConfig) Swap(i, j int) { r[i], r[j] = r[j], r[i] }
func (r byPortConfig) Less(i, j int) bool {
// We convert PortConfig into `port/protocol`, e.g., `80/tcp`
// In updatePorts we already filter out with map so there is duplicate entries
return portConfigToString(&r[i]) < portConfigToString(&r[j])
}
func portConfigToString(portConfig *swarm.PortConfig) string {
protocol := portConfig.Protocol
if protocol == "" {
protocol = "tcp"
}
return fmt.Sprintf("%v/%s", portConfig.PublishedPort, protocol)
}
func updatePorts(flags *pflag.FlagSet, portConfig *[]swarm.PortConfig) error {
// The key of the map is `port/protocol`, e.g., `80/tcp`
portSet := map[string]swarm.PortConfig{}
// Check to see if there are any conflict in flags.
if flags.Changed(flagPublishAdd) {
values := flags.Lookup(flagPublishAdd).Value.(*opts.ListOpts).GetAll()
ports, portBindings, _ := nat.ParsePortSpecs(values)
for port := range ports {
newConfigs := convertPortToPortConfig(port, portBindings)
for _, entry := range newConfigs {
if v, ok := portSet[portConfigToString(&entry)]; ok && v != entry {
return fmt.Errorf("conflicting port mapping between %v:%v/%s and %v:%v/%s", entry.PublishedPort, entry.TargetPort, entry.Protocol, v.PublishedPort, v.TargetPort, v.Protocol)
}
portSet[portConfigToString(&entry)] = entry
}
}
}
// Override previous PortConfig in service if there is any duplicate
for _, entry := range *portConfig {
if _, ok := portSet[portConfigToString(&entry)]; !ok {
portSet[portConfigToString(&entry)] = entry
}
}
toRemove := flags.Lookup(flagPublishRemove).Value.(*opts.ListOpts).GetAll()
newPorts := []swarm.PortConfig{}
portLoop:
for _, port := range portSet {
for _, rawTargetPort := range toRemove {
targetPort := nat.Port(rawTargetPort)
if equalPort(targetPort, port) {
continue portLoop
}
}
newPorts = append(newPorts, port)
}
// Sort the PortConfig to avoid unnecessary updates
sort.Sort(byPortConfig(newPorts))
*portConfig = newPorts
return nil
}
func equalPort(targetPort nat.Port, port swarm.PortConfig) bool {
return (string(port.Protocol) == targetPort.Proto() &&
port.TargetPort == uint32(targetPort.Int()))
}
func updateReplicas(flags *pflag.FlagSet, serviceMode *swarm.ServiceMode) error {
if !flags.Changed(flagReplicas) {
return nil
}
if serviceMode == nil || serviceMode.Replicated == nil {
return fmt.Errorf("replicas can only be used with replicated mode")
}
serviceMode.Replicated.Replicas = flags.Lookup(flagReplicas).Value.(*Uint64Opt).Value()
return nil
}
// updateLogDriver updates the log driver only if the log driver flag is set.
// All options will be replaced with those provided on the command line.
func updateLogDriver(flags *pflag.FlagSet, taskTemplate *swarm.TaskSpec) error {
if !flags.Changed(flagLogDriver) {
return nil
}
name, err := flags.GetString(flagLogDriver)
if err != nil {
return err
}
if name == "" {
return nil
}
taskTemplate.LogDriver = &swarm.Driver{
Name: name,
Options: runconfigopts.ConvertKVStringsToMap(flags.Lookup(flagLogOpt).Value.(*opts.ListOpts).GetAll()),
}
return nil
}