terraform/helper/resource/state.go

package resource

import (
	"log"
	"time"
)

var refreshGracePeriod = 30 * time.Second

// StateRefreshFunc is a function type used for StateChangeConf that is
// responsible for refreshing the item being watched for a state change.
//
// It returns three results. `result` is any object that will be returned
// as the final object after waiting for state change. This allows you to
// return the final updated object, for example an EC2 instance after refreshing
// it.
//
// `state` is the latest state of that object. And `err` is any error that
// may have happened while refreshing the state.
type StateRefreshFunc func() (result interface{}, state string, err error)

// StateChangeConf is the configuration struct used for `WaitForState`.
type StateChangeConf struct {
	Delay          time.Duration    // Wait this time before starting checks
	Pending        []string         // States that are "allowed" and will continue trying
	Refresh        StateRefreshFunc // Refreshes the current state
	Target         []string         // Target state
	Timeout        time.Duration    // The amount of time to wait before timeout
	MinTimeout     time.Duration    // Smallest time to wait before refreshes
	PollInterval   time.Duration    // Override MinTimeout/backoff and only poll this often
	NotFoundChecks int              // Number of times to allow not found

	// This is to work around inconsistent APIs
	ContinuousTargetOccurence int // Number of times the Target state has to occur continuously
}

// WaitForState watches an object and waits for it to achieve the state
// specified in the configuration using the specified Refresh() func,
// waiting the number of seconds specified in the timeout configuration.
//
// If the Refresh function returns a error, exit immediately with that error.
//
// If the Refresh function returns a state other than the Target state or one
// listed in Pending, return immediately with an error.
//
// If the Timeout is exceeded before reaching the Target state, return an
// error.
//
// Otherwise, result the result of the first call to the Refresh function to
// reach the target state.
func (conf *StateChangeConf) WaitForState() (interface{}, error) {
	log.Printf("[DEBUG] Waiting for state to become: %s", conf.Target)

	notfoundTick := 0
	targetOccurence := 0

	// Set a default for times to check for not found
	if conf.NotFoundChecks == 0 {
		conf.NotFoundChecks = 20
	}

	if conf.ContinuousTargetOccurence == 0 {
		conf.ContinuousTargetOccurence = 1
	}

	type Result struct {
		Result interface{}
		State  string
		Error  error
		Done   bool
	}

	// Read every result from the refresh loop, waiting for a positive result.Done.
	resCh := make(chan Result, 1)
	// cancellation channel for the refresh loop
	cancelCh := make(chan struct{})

	result := Result{}

	go func() {
		defer close(resCh)

		time.Sleep(conf.Delay)

		// start with 0 delay for the first loop
		var wait time.Duration

		for {
			// store the last result
			resCh <- result

			// wait and watch for cancellation
			select {
			case <-cancelCh:
				return
			case <-time.After(wait):
				// first round had no wait
				if wait == 0 {
					wait = 100 * time.Millisecond
				}
			}

			res, currentState, err := conf.Refresh()
			result = Result{
				Result: res,
				State:  currentState,
				Error:  err,
			}

			if err != nil {
				resCh <- result
				return
			}

			// If we're waiting for the absence of a thing, then return
			if res == nil && len(conf.Target) == 0 {
				targetOccurence++
				if conf.ContinuousTargetOccurence == targetOccurence {
					result.Done = true
					resCh <- result
					return
				}
				continue
			}

			if res == nil {
				// If we didn't find the resource, check if we have been
				// not finding it for awhile, and if so, report an error.
				notfoundTick++
				if notfoundTick > conf.NotFoundChecks {
					result.Error = &NotFoundError{
						LastError: err,
						Retries:   notfoundTick,
					}
					resCh <- result
					return
				}
			} else {
				// Reset the counter for when a resource isn't found
				notfoundTick = 0
				found := false

				for _, allowed := range conf.Target {
					if currentState == allowed {
						found = true
						targetOccurence++
						if conf.ContinuousTargetOccurence == targetOccurence {
							result.Done = true
							resCh <- result
							return
						}
						continue
					}
				}

				for _, allowed := range conf.Pending {
					if currentState == allowed {
						found = true
						targetOccurence = 0
						break
					}
				}

				if !found && len(conf.Pending) > 0 {
					result.Error = &UnexpectedStateError{
						LastError:     err,
						State:         result.State,
						ExpectedState: conf.Target,
					}
					resCh <- result
					return
				}
			}

			// Wait between refreshes using exponential backoff, except when
			// waiting for the target state to reoccur.
			if targetOccurence == 0 {
				wait *= 2
			}

			// If a poll interval has been specified, choose that interval.
			// Otherwise bound the default value.
			if conf.PollInterval > 0 && conf.PollInterval < 180*time.Second {
				wait = conf.PollInterval
			} else {
				if wait < conf.MinTimeout {
					wait = conf.MinTimeout
				} else if wait > 10*time.Second {
					wait = 10 * time.Second
				}
			}

			log.Printf("[TRACE] Waiting %s before next try", wait)
		}
	}()

	// store the last value result from the refresh loop
	lastResult := Result{}

	timeout := time.After(conf.Timeout)
	for {
		select {
		case r, ok := <-resCh:
			// channel closed, so return the last result
			if !ok {
				return lastResult.Result, lastResult.Error
			}

			// we reached the intended state
			if r.Done {
				return r.Result, r.Error
			}

			// still waiting, store the last result
			lastResult = r

		case <-timeout:
			log.Printf("[WARN] WaitForState timeout after %s", conf.Timeout)
			log.Printf("[WARN] WaitForState starting %s refresh grace period", refreshGracePeriod)

			// cancel the goroutine and start our grace period timer
			close(cancelCh)
			timeout := time.After(refreshGracePeriod)

			// we need a for loop and a label to break on, because we may have
			// an extra response value to read, but still want to wait for the
			// channel to close.
		forSelect:
			for {
				select {
				case r, ok := <-resCh:
					if r.Done {
						// the last refresh loop reached the desired state
						return r.Result, r.Error
					}

					if !ok {
						// the goroutine returned
						break forSelect
					}

					// target state not reached, save the result for the
					// TimeoutError and wait for the channel to close
					lastResult = r
				case <-timeout:
					log.Println("[ERROR] WaitForState exceeded refresh grace period")
					break forSelect
				}
			}

			return nil, &TimeoutError{
				LastError:     lastResult.Error,
				LastState:     lastResult.State,
				Timeout:       conf.Timeout,
				ExpectedState: conf.Target,
			}
		}
	}
}