feat: ask whether to reset already initialised machine on 'machine init/add'

cmd/uncloud/machine/add.go

@@ -41,7 +41,7 @@ func NewAddCommand() *cobra.Command {
 			if err != nil {
 				return fmt.Errorf("parse remote machine: %w", err)
 			}
-			remoteMachine := cli.RemoteMachine{
+			remoteMachine := &cli.RemoteMachine{
 				User:    user,
 				Host:    host,
 				Port:    port,
@@ -62,8 +62,9 @@ func NewAddCommand() *cobra.Command {
 			fmt.Sprintf("blank '' or '%s' to disable ingress on this machine, or specify an IP address.", PublicIPNone),
 	)
 	cmd.Flags().StringVarP(
-		&opts.sshKey, "ssh-key", "i", "~/.ssh/id_ed25519",
-		"Path to SSH private key for remote login (if not already added to SSH agent).",
+		&opts.sshKey, "ssh-key", "i", "",
+		fmt.Sprintf("Path to SSH private key for remote login (if not already added to SSH agent). (default %q)",
+			cli.DefaultSSHKeyPath),
 	)
 	cmd.Flags().StringVar(
 		&opts.version, "version", "latest",
@@ -77,7 +78,7 @@ func NewAddCommand() *cobra.Command {
 	return cmd
 }
 
-func add(ctx context.Context, uncli *cli.CLI, remoteMachine cli.RemoteMachine, opts addOptions) error {
+func add(ctx context.Context, uncli *cli.CLI, remoteMachine *cli.RemoteMachine, opts addOptions) error {
 	var publicIP *netip.Addr
 	switch opts.publicIP {
 	case "auto":

cmd/uncloud/machine/init.go

@@ -80,8 +80,9 @@ func NewInitCommand() *cobra.Command {
 			fmt.Sprintf("blank '' or '%s' to disable ingress on this machine, or specify an IP address.", PublicIPNone),
 	)
 	cmd.Flags().StringVarP(
-		&opts.sshKey, "ssh-key", "i", "~/.ssh/id_ed25519",
-		"Path to SSH private key for remote login (if not already added to SSH agent).",
+		&opts.sshKey, "ssh-key", "i", "",
+		fmt.Sprintf("Path to SSH private key for remote login (if not already added to SSH agent). (default %q)",
+			cli.DefaultSSHKeyPath),
 	)
 	cmd.Flags().StringVar(
 		&opts.version, "version", "latest",

cmd/uncloud/machine/rm.go

@@ -90,11 +90,13 @@ func remove(ctx context.Context, uncli *cli.CLI, nameOrID string, opts removeOpt
 
 	reset := !opts.noReset
 	var containers []api.ServiceContainer
+	reachable := false
 	if reset {
 		// Check if the machine is up and has service containers.
 		listOpts := container.ListOptions{All: true}
 		machineContainers, err := client.Docker.ListServiceContainers(mctx, "", listOpts)
 		if err == nil {
+			reachable = true
 			containers = machineContainers[0].Containers
 			if len(containers) > 0 {
 				plural := ""
@@ -104,7 +106,7 @@ func remove(ctx context.Context, uncli *cli.CLI, nameOrID string, opts removeOpt
 				fmt.Printf("Found %d service container%s on machine '%s':\n", len(containers), plural, m.Name)
 				fmt.Println(formatContainerTree(containers))
 				fmt.Println()
-				fmt.Println("This will remove all service containers on the machine, remove it from the cluster, " +
+				fmt.Println("This will remove all service containers from the machine, remove it from the cluster, " +
 					"and reset it to the uninitialised state.")
 			} else {
 				fmt.Printf("No service containers found on machine '%s'.\n", m.Name)
@@ -129,8 +131,7 @@ func remove(ctx context.Context, uncli *cli.CLI, nameOrID string, opts removeOpt
 		}
 	}
 
-	if reset {
-		if len(containers) > 0 {
+	if reset && len(containers) > 0 {
 		err = progress.RunWithTitle(ctx, func(ctx context.Context) error {
 			return removeContainers(ctx, client, containers)
 		}, uncli.ProgressOut(), "Removing containers")
@@ -139,14 +140,13 @@ func remove(ctx context.Context, uncli *cli.CLI, nameOrID string, opts removeOpt
 		}
 		fmt.Println()
 	}
-	}
 
 	if _, err = client.RemoveMachine(ctx, &pb.RemoveMachineRequest{Id: m.Id}); err != nil {
 		return fmt.Errorf("remove machine from cluster: %w", err)
 	}
 	fmt.Printf("Machine '%s' removed from the cluster.\n", m.Name)
 
-	if reset {
+	if reset && reachable {
 		_, err = client.MachineClient.Reset(mctx, &pb.ResetRequest{})
 		if err != nil {
 			fmt.Printf("WARNING: Failed to reset machine: %v\n", err)

internal/cli/cli.go

@@ -6,8 +6,8 @@ import (
 	"fmt"
 	"net/netip"
 	"os"
+	"slices"
 
-	"github.com/charmbracelet/huh"
 	"github.com/docker/cli/cli/streams"
 	"github.com/psviderski/uncloud/internal/cli/config"
 	"github.com/psviderski/uncloud/internal/fs"
@@ -22,7 +22,12 @@ import (
 	"google.golang.org/protobuf/types/known/emptypb"
 )
 
-const defaultContextName = "default"
+const (
+	// DefaultSSHKeyPath is the fallback location for the SSH private key when provisioning remote machines.
+	// Used when no key is explicitly provided and SSH agent authentication fails.
+	DefaultSSHKeyPath  = "~/.ssh/id_ed25519"
+	defaultContextName = "default"
+)
 
 type CLI struct {
 	Config *config.Config
@@ -173,7 +178,7 @@ func (cli *CLI) initRemoteMachine(ctx context.Context, opts InitClusterOptions)
 		return nil, fmt.Errorf("cluster context '%s' already exists", contextName)
 	}
 
-	machineClient, err := cli.provisionRemoteMachine(ctx, *opts.RemoteMachine, opts.Version)
+	machineClient, err := provisionRemoteMachine(ctx, opts.RemoteMachine, opts.Version)
 	if err != nil {
 		return nil, err
 	}
@@ -190,7 +195,7 @@ func (cli *CLI) initRemoteMachine(ctx context.Context, opts InitClusterOptions)
 		return nil, fmt.Errorf("inspect machine: %w", err)
 	}
 	if minfo.Id != "" {
-		if err = cli.promptResetMachine(); err != nil {
+		if err = promptResetMachine(ctx, machineClient.MachineClient); err != nil {
 			return nil, err
 		}
 	}
@@ -249,7 +254,7 @@ type AddMachineOptions struct {
 	Context       string
 	MachineName   string
 	PublicIP      *netip.Addr
-	RemoteMachine RemoteMachine
+	RemoteMachine *RemoteMachine
 	Version       string
 }
 
@@ -272,7 +277,7 @@ func (cli *CLI) AddMachine(ctx context.Context, opts AddMachineOptions) (*client
 		}
 	}()
 
-	machineClient, err := cli.provisionRemoteMachine(ctx, opts.RemoteMachine, opts.Version)
+	machineClient, err := provisionRemoteMachine(ctx, opts.RemoteMachine, opts.Version)
 	if err != nil {
 		return nil, nil, err
 	}
@@ -288,7 +293,18 @@ func (cli *CLI) AddMachine(ctx context.Context, opts AddMachineOptions) (*client
 		return nil, nil, fmt.Errorf("inspect machine: %w", err)
 	}
 	if minfo.Id != "" {
-		if err = cli.promptResetMachine(); err != nil {
+		// Check if the machine is already a member of this cluster.
+		machines, err := c.ListMachines(ctx, nil)
+		if err != nil {
+			return nil, nil, fmt.Errorf("list cluster machines: %w", err)
+		}
+		if slices.ContainsFunc(machines, func(m *pb.MachineMember) bool {
+			return m.Machine.Id == minfo.Id
+		}) {
+			return nil, nil, fmt.Errorf("machine is already a member of this cluster (%s)", minfo.Name)
+		}
+
+		if err = promptResetMachine(ctx, machineClient.MachineClient); err != nil {
 			return nil, nil, err
 		}
 	}
@@ -339,7 +355,7 @@ func (cli *CLI) AddMachine(ctx context.Context, opts AddMachineOptions) (*client
 		return nil, nil, fmt.Errorf("add machine to cluster (context '%s'): %w", contextName, err)
 	}
 
-	// List other machines in the cluster to include them in the join request.
+	// Get the most up-to-date list of other machines in the cluster to include them in the join request.
 	machines, err := c.ListMachines(ctx, nil)
 	if err != nil {
 		return nil, nil, fmt.Errorf("list cluster machines: %w", err)
@@ -382,11 +398,20 @@ func (cli *CLI) AddMachine(ctx context.Context, opts AddMachineOptions) (*client
 // provisionRemoteMachine installs the Uncloud daemon and dependencies on the remote machine over SSH and returns
 // a machine API client to interact with the machine. The client should be closed after use by the caller.
 // The version parameter specifies the version of the Uncloud daemon to install. If empty, the latest version is used.
-func (cli *CLI) provisionRemoteMachine(
-	ctx context.Context, remoteMachine RemoteMachine, version string,
+// The remoteMachine.SSHKeyPath could be updated to the default SSH key path if it is not set and the SSH agent
+// authentication fails.
+func provisionRemoteMachine(
+	ctx context.Context, remoteMachine *RemoteMachine, version string,
 ) (*client.Client, error) {
 	// Provision the remote machine by installing the Uncloud daemon and dependencies over SSH.
 	sshClient, err := sshexec.Connect(remoteMachine.User, remoteMachine.Host, remoteMachine.Port, remoteMachine.KeyPath)
+	// If the SSH connection using SSH agent fails and no key path is provided, try to use the default SSH key.
+	if err != nil && remoteMachine.KeyPath == "" {
+		remoteMachine.KeyPath = DefaultSSHKeyPath
+		sshClient, err = sshexec.Connect(
+			remoteMachine.User, remoteMachine.Host, remoteMachine.Port, remoteMachine.KeyPath,
+		)
+	}
 	if err != nil {
 		return nil, fmt.Errorf(
 			"SSH login to remote machine %s: %w",
@@ -420,31 +445,6 @@ func (cli *CLI) provisionRemoteMachine(
 	return machineClient, nil
 }
 
-func (cli *CLI) promptResetMachine() error {
-	var confirm bool
-	form := huh.NewForm(
-		huh.NewGroup(
-			huh.NewConfirm().
-				Title(
-					"The remote machine is already initialised as a cluster member. Do you want to reset it first?",
-				).
-				Affirmative("Yes!").
-				Negative("No").
-				Value(&confirm),
-		),
-	).WithAccessible(true)
-	if err := form.Run(); err != nil {
-		return fmt.Errorf("prompt user to confirm: %w", err)
-	}
-
-	if !confirm {
-		return fmt.Errorf("remote machine is already initialised as a cluster member")
-	}
-	// TODO: implement resetting the remote machine.
-	return fmt.Errorf("resetting the remote machine is not implemented yet. " +
-		"Please manually run 'uncloud-uninstall' on the remote machine to fully uninstall Uncloud from it")
-}
-
 // ProgressOut returns an output stream for progress writer.
 func (cli *CLI) ProgressOut() *streams.Out {
 	return streams.NewOut(os.Stdout)

internal/cli/machine.go

@@ -5,12 +5,20 @@ import (
 	"fmt"
 	"os"
 	"strings"
+	"time"
 
+	"github.com/cenkalti/backoff/v4"
+	"github.com/charmbracelet/huh"
+	"github.com/psviderski/uncloud/internal/machine/api/pb"
 	"github.com/psviderski/uncloud/internal/sshexec"
+	"google.golang.org/protobuf/types/known/emptypb"
 )
 
-// TODO: support pinning the script version to the CLI version.
-const installScriptURL = "https://raw.githubusercontent.com/psviderski/uncloud/refs/heads/main/scripts/install.sh"
+const (
+	// TODO: support pinning the script version to the CLI version.
+	installScriptURL = "https://raw.githubusercontent.com/psviderski/uncloud/refs/heads/main/scripts/install.sh"
+	rootUser         = "root"
+)
 
 type RemoteMachine struct {
 	User    string
@@ -24,7 +32,7 @@ func installCmd(user string, version string) string {
 	var env []string
 
 	// Add the SSH user (non-root) to the uncloud group to allow access to the Uncloud daemon unix socket.
-	if user != "root" {
+	if user != rootUser {
 		sudoPrefix = "sudo"
 		env = append(env, "UNCLOUD_GROUP_ADD_USER="+sshexec.Quote(user))
 	}
@@ -46,6 +54,26 @@ func provisionMachine(ctx context.Context, exec sshexec.Executor, version string
 		return fmt.Errorf("run whoami: %w", err)
 	}
 
+	if user != rootUser {
+		// 'sudo -n' is not used because it fails with 'sudo: a password is required' when the user has no password
+		// in /etc/shadow even though it may have valid sudo access.
+		out, err := exec.Run(ctx, "sudo true")
+		if err != nil {
+			if strings.Contains(out, "password is required") {
+				return fmt.Errorf(
+					"user '%[1]s' requires a password for sudo, but Uncloud needs passwordless sudo or root access "+
+						"to install and configure the uncloudd daemon on the remote machine.\n\n"+
+						"Possible solutions:\n"+
+						"1. Use root user or a user with passwordless sudo instead.\n"+
+						"2. Configure passwordless sudo for the user '%[1]s' by running on the remote machine:\n"+
+						"   echo '%[1]s ALL=(ALL) NOPASSWD:ALL' | sudo tee /etc/sudoers.d/%[1]s",
+					user)
+			}
+			return fmt.Errorf("sudo command failed for user '%s': %w. "+
+				"Please ensure the user has sudo privileges or use root user instead", user, err)
+		}
+	}
+
 	cmd := installCmd(user, version)
 
 	fmt.Println("Downloading Uncloud install script:", installScriptURL)
@@ -56,3 +84,56 @@ func provisionMachine(ctx context.Context, exec sshexec.Executor, version string
 	}
 	return nil
 }
+
+func promptResetMachine(ctx context.Context, machineClient pb.MachineClient) error {
+	var confirm bool
+	form := huh.NewForm(
+		huh.NewGroup(
+			huh.NewConfirm().
+				Title(
+					"The remote machine is already initialised as a cluster member. Do you want to reset it first?\n" +
+						"This will:\n" +
+						"- Remove all service containers from the machine\n" +
+						"- Reset the machine to the uninitialised state",
+				).
+				Affirmative("Yes!").
+				Negative("No").
+				Value(&confirm),
+		),
+	).WithAccessible(true)
+	if err := form.Run(); err != nil {
+		return fmt.Errorf("prompt user to confirm: %w", err)
+	}
+
+	if !confirm {
+		return fmt.Errorf("remote machine is already initialised as a cluster member")
+	}
+
+	if _, err := machineClient.Reset(ctx, &pb.ResetRequest{}); err != nil {
+		return fmt.Errorf("reset remote machine: %w. You can also manually run 'uncloud-uninstall' "+
+			"on the remote machine to fully uninstall Uncloud from it", err)
+	}
+	fmt.Println("Resetting the remote machine...")
+	if err := waitMachineReady(ctx, machineClient, 1*time.Minute); err != nil {
+		return fmt.Errorf("wait for machine to be ready after reset: %w", err)
+	}
+
+	return nil
+}
+
+// waitMachineReady waits for the machine to be ready to serve requests.
+func waitMachineReady(ctx context.Context, machineClient pb.MachineClient, timeout time.Duration) error {
+	boff := backoff.WithContext(backoff.NewExponentialBackOff(
+		backoff.WithMaxInterval(1*time.Second),
+		backoff.WithMaxElapsedTime(timeout),
+	), ctx)
+
+	inspect := func() error {
+		_, err := machineClient.Inspect(ctx, &emptypb.Empty{})
+		if err != nil {
+			return fmt.Errorf("inspect machine: %w", err)
+		}
+		return nil
+	}
+	return backoff.Retry(inspect, boff)
+}

internal/machine/machine.go

@@ -446,10 +446,7 @@ func (m *Machine) Run(ctx context.Context) error {
 		slog.Info("Local API proxy server stopped.")
 
 		// Clean up the machine data and resources if the machine shutdown was initiated by a reset.
-		m.mu.RLock()
-		resetting := m.resetting
-		m.mu.RUnlock()
-		if resetting {
+		if m.resetting {
 			slog.Info("Cleaning up machine data and resources.")
 			if err = m.cleanup(); err != nil {
 				slog.Error("Failed to clean up machine data and resources.", "err", err)

pkg/client/connector/ssh.go

@@ -63,7 +63,7 @@ func (c *SSHConnector) Connect(ctx context.Context) (*grpc.ClientConn, error) {
 				conn, dErr := c.client.DialContext(ctx, "unix", addr)
 				if dErr != nil {
 					return nil, fmt.Errorf(
-						"connect to machine API socket '%s' through SSH tunnel (is the Uncloud daemon running "+
+						"connect to machine API socket '%s' through SSH tunnel (is uncloud.service running "+
 							"on the remote machine and does the SSH user '%s' have permissions to access the socket?):"+
 							" %w",
 						addr, c.client.User(), dErr,

website/docs/2-getting-started/2-deploy-demo-app.md

@@ -9,8 +9,8 @@ infrastructure with secure internet access.
 Before you begin, you'll need:
 
 - **Uncloud CLI** [installed](1-install-cli.md) on your local machine
-- A **Ubuntu or Debian server** with **public IP address** and **SSH access** (as `root` or a user with `sudo`
-  privileges) using a **private key**.
+- A **Ubuntu or Debian server** with **public IP address** and **SSH access** using a **private key** (as `root` or a
+  user with **passwordless** `sudo` privileges).
 
 :::tip Need a server?
 
@@ -273,9 +273,11 @@ Add a CNAME record `excalidraw.example.com` in your DNS provider (Cloudflare, Na
 
 :::info note
 
-These instructions set up your own domain _in addition to_ uncloud's managed DNS service.
+These instructions set up your own domain **in addition to** the Uncloud managed DNS name
+`excalidraw.7za6s7.cluster.uncloud.run`.
 
-If you want to avoid the managed service altogether, add `--no-dns` to your `uc machine init` command, and point an `A`-type DNS record to your server(s)'s IP(s).
+If you want to avoid the managed service altogether, add `--no-dns` to your `uc machine init` command, and point an A
+DNS record to your server(s)'s IP(s).
 
 :::