builder/amazon: Extend timeout and allow user override [GH-1533]

- help resolve https://github.com/mitchellh/packer/issues/1533
  (although timeouts are always ultimately useless in a distributed
  system!)
- makes packer no more idempotent or janitorial than before
- derive maximum number of ticks from timeout
- default timeout to 300s (5m) to cater for global AMI copying
- allow user to override with AWS_TIMEOUT_SECONDS environment variable

CHANGELOG.md

@@ -11,6 +11,7 @@ BUG FIXES:
   * builder/vmware: Wait some time so that post-VMX update aren't
       overwritten. [GH-1504]
   * builder/vmware-vmx: Fix issue with order of boot command support [GH-1492]
+  * builder/amazon: Extend timeout and allow user override [GH-1533]
   * builder/parallels: Ignore 'The fdd0 device does not exist' [GH-1501]
   * builder/parallels: Rely on Cleanup functions to detach devices [GH-1502]
   * builder/parallels: Create VM without hdd and then add it later [GH-1548]

builder/amazon/common/state.go

@@ -7,6 +7,8 @@ import (
 	"github.com/mitchellh/multistep"
 	"log"
 	"net"
+	"os"
+	"strconv"
 	"time"
 )
 
@@ -122,6 +124,8 @@ func SpotRequestStateRefreshFunc(conn *ec2.EC2, spotRequestId string) StateRefre
 func WaitForState(conf *StateChangeConf) (i interface{}, err error) {
 	log.Printf("Waiting for state to become: %s", conf.Target)
 
+	sleepSeconds := 2
+	maxTicks := int(TimeoutSeconds() / sleepSeconds) + 1
 	notfoundTick := 0
 
 	for {
@@ -135,7 +139,7 @@ func WaitForState(conf *StateChangeConf) (i interface{}, err error) {
 			// If we didn't find the resource, check if we have been
 			// not finding it for awhile, and if so, report an error.
 			notfoundTick += 1
-			if notfoundTick > 20 {
+			if notfoundTick > maxTicks {
 				return nil, errors.New("couldn't find resource")
 			}
 		} else {
@@ -166,7 +170,7 @@ func WaitForState(conf *StateChangeConf) (i interface{}, err error) {
 			}
 		}
 
-		time.Sleep(2 * time.Second)
+		time.Sleep(time.Duration(sleepSeconds) * time.Second)
 	}
 
 	return
@@ -179,3 +183,23 @@ func isTransientNetworkError(err error) bool {
 
 	return false
 }
+
+// Returns 300 seconds (5 minutes) by default
+// Some AWS operations, like copying an AMI to a distant region, take a very long time
+// Allow user to override with AWS_TIMEOUT_SECONDS environment variable
+func TimeoutSeconds() (seconds int) {
+	seconds = 300
+
+	override := os.Getenv("AWS_TIMEOUT_SECONDS")
+	if override != "" {
+		n, err := strconv.Atoi(override)
+		if err != nil {
+			log.Printf("Invalid timeout seconds '%s', using default", override)
+		} else {
+			seconds = n
+		}
+	}
+
+	log.Printf("Allowing %ds to complete (change with AWS_TIMEOUT_SECONDS)", seconds)
+	return seconds
+}