fly deploy: concurrently acquire/release leases, update flaps.Wait() …

…to better handle timeouts

api/machine_types.go

@@ -6,11 +6,14 @@ import (
 )
 
 const MachineConfigMetadataKeyFlyPlatformVersion = "fly_platform_version"
-const MachineConfigMetadataKeyFlyRelease = "fly_release"
+const MachineConfigMetadataKeyFlyReleaseId = "fly_release_id"
+const MachineConfigMetadataKeyFlyReleaseVersion = "fly_release_version"
 const MachineFlyPlatformVersion2 = "v2"
 const MachineProcessGroupApp = "app"
 const MachineProcessGroupReleaseCommand = "release_command"
 const MachineStateDestroyed = "destroyed"
+const MachineStateStarted = "started"
+const MachineStateStopped = "stopped"
 
 type Machine struct {
 	ID       string          `json:"id"`
@@ -236,12 +239,16 @@ type DNSConfig struct {
 }
 
 type MachineLease struct {
-	Status string `json:"status"`
-	Data   struct {
-		Nonce     string `json:"nonce"`
-		ExpiresAt int64  `json:"expires_at"`
-		Owner     string `json:"owner"`
-	}
+	Status  string            `json:"status"`
+	Data    *MachineLeaseData `json:"data,omitempty"`
+	Message string            `json:"message,omitempty"`
+	Code    string            `json:"code,omitempty"`
+}
+
+type MachineLeaseData struct {
+	Nonce     string `json:"nonce"`
+	ExpiresAt int64  `json:"expires_at"`
+	Owner     string `json:"owner"`
 }
 
 type MachineStartResponse struct {

flaps/flaps.go

@@ -10,7 +10,9 @@ import (
 	"net"
 	"net/http"
 	"strings"
+	"time"
 
+	"github.com/google/go-querystring/query"
 	"github.com/samber/lo"
 
 	"github.com/superfly/flyctl/agent"
@@ -119,26 +121,40 @@ func (f *Client) Start(ctx context.Context, machineID string) (*api.MachineStart
 	return out, nil
 }
 
-func (f *Client) Wait(ctx context.Context, machine *api.Machine, state string) (err error) {
-	waitEndpoint := fmt.Sprintf("/%s/wait", machine.ID)
+type waitQuerystring struct {
+	InstanceId     string `url:"instance_id,omitempty"`
+	TimeoutSeconds int    `url:"timeout,omitempty"`
+	State          string `url:"state,omitempty"`
+}
 
-	version := machine.InstanceID
+const proxyTimeoutThreshold = 60 * time.Second
 
+func (f *Client) Wait(ctx context.Context, machine *api.Machine, state string, timeout time.Duration) (err error) {
+	waitEndpoint := fmt.Sprintf("/%s/wait", machine.ID)
+	if state == "" {
+		state = "started"
+	}
+	version := machine.InstanceID
 	if machine.Version != "" {
 		version = machine.Version
 	}
-	if version != "" {
-		waitEndpoint += fmt.Sprintf("?instance_id=%s&timeout=30", version)
-	} else {
-		waitEndpoint += "?timeout=30"
+	if timeout > proxyTimeoutThreshold {
+		timeout = proxyTimeoutThreshold
 	}
-
-	if state == "" {
-		state = "started"
+	if timeout < 1*time.Second {
+		timeout = 1 * time.Second
 	}
-
-	waitEndpoint += fmt.Sprintf("&state=%s", state)
-
+	timeoutSeconds := int(timeout.Seconds())
+	waitQs := waitQuerystring{
+		InstanceId:     version,
+		TimeoutSeconds: timeoutSeconds,
+		State:          state,
+	}
+	qsVals, err := query.Values(waitQs)
+	if err != nil {
+		return fmt.Errorf("error making query string for wait request: %w", err)
+	}
+	waitEndpoint += fmt.Sprintf("?%s", qsVals.Encode())
 	if err := f.sendRequest(ctx, http.MethodGet, waitEndpoint, nil, nil, nil); err != nil {
 		return fmt.Errorf("failed to wait for VM %s in %s state: %w", machine.ID, state, err)
 	}