felhom-agent/internal/config/config.go

// Package config loads the felhom-agent configuration the proxmox layer needs.
//
// Format: a JSON file (stdlib-only — no YAML dep, consistent with the agent's
// "pure stdlib" constraint), with per-field environment overrides. Secrets (the
// API token) are never logged; see Config.Redacted.
//
// OPEN item (noted in the slice reply): the controller/hub use YAML; if matching
// that house style is preferred over the zero-dependency constraint, the loader
// can swap to yaml.v3 without touching call sites.
package config

import (
	"encoding/json"
	"fmt"
	"net"
	"net/url"
	"os"
	"strconv"
	"strings"
	"time"
)

// Config is the agent configuration.
type Config struct {
	Proxmox    ProxmoxConfig    `json:"proxmox"`
	Privileged PrivilegedConfig `json:"privileged"`
	Authz      AuthzConfig      `json:"authz"`
	Hub        HubConfig        `json:"hub"`
	Storage    StorageConfig    `json:"storage"`
	Backup     BackupConfig     `json:"backup"`
	Escrow     EscrowConfig     `json:"escrow"`
	LocalAPI   LocalAPIConfig   `json:"local_api"`
	LogLevel   string           `json:"log_level"` // debug|info|warn|error (default info)
}

// LocalAPIConfig configures the per-guest local API server (doc 03 §6, slice 8A). The
// controller (inside its LXC) reaches the agent over the local bridge; the agent is the
// per-guest authorization gate — it maps a per-guest bearer token → VMID and authorizes
// every call against THAT guest only (self-scoped; never a caller-supplied id). Disabled
// unless Enable is set AND ListenAddr is non-empty (so a host that doesn't yet provision
// controllers runs the daemon without it).
//
// Defense-in-depth (spike gotcha 5): ListenAddr should be the host's BRIDGE IP (not
// 0.0.0.0), and a host firewall rule should limit the port to the guest bridge subnet
// (configs/felhom-localapi-firewall.example). The per-guest token remains the gate; the
// bind + firewall narrow exposure but are not the authorization.
type LocalAPIConfig struct {
	Enable     bool   `json:"enable"`
	ListenAddr string `json:"listen_addr"` // bridge IP:port, e.g. "192.168.0.162:8443"
	// CertFile/KeyFile hold the agent's self-signed leaf served to controllers. Generated
	// (persisted) on first start if absent, so the leaf SHA-256 fingerprint — baked into each
	// guest's bootstrap for pinning — is STABLE across agent restarts.
	CertFile string `json:"cert_file"` // default <token_store_dir>/local-api.crt
	KeyFile  string `json:"key_file"`  // default <token_store_dir>/local-api.key
	// TokenStore is the durable, hashed token→guest map (only a HASH of each token is
	// persisted; the plaintext exists transiently at mint→write-to-mount, then is discarded).
	TokenStore string `json:"token_store"` // default /var/lib/felhom-agent/local-tokens.log
}

// Default local-API file locations (under the agent's state dir).
const (
	defaultLocalAPITokenStore = "/var/lib/felhom-agent/local-tokens.log"
	defaultLocalAPICert       = "/var/lib/felhom-agent/local-api.crt"
	defaultLocalAPIKey        = "/var/lib/felhom-agent/local-api.key"
)

// Enabled reports whether the local-API server should run.
func (l LocalAPIConfig) Enabled() bool {
	return l.Enable && strings.TrimSpace(l.ListenAddr) != ""
}

// TokenStorePath returns the configured token-store path (default applied).
func (l LocalAPIConfig) TokenStorePath() string {
	if l.TokenStore != "" {
		return l.TokenStore
	}
	return defaultLocalAPITokenStore
}

// CertPath/KeyPath return the configured leaf cert/key paths (defaults applied).
func (l LocalAPIConfig) CertPath() string {
	if l.CertFile != "" {
		return l.CertFile
	}
	return defaultLocalAPICert
}

func (l LocalAPIConfig) KeyPath() string {
	if l.KeyFile != "" {
		return l.KeyFile
	}
	return defaultLocalAPIKey
}

// Validate checks the local-API config is usable when enabled.
func (l LocalAPIConfig) Validate() error {
	if !l.Enable {
		return nil
	}
	if strings.TrimSpace(l.ListenAddr) == "" {
		return fmt.Errorf("config: local_api.listen_addr is required when local_api.enable is set (use the host bridge IP:port, e.g. 192.168.0.162:8443)")
	}
	if _, _, err := net.SplitHostPort(l.ListenAddr); err != nil {
		return fmt.Errorf("config: local_api.listen_addr %q is not host:port: %w", l.ListenAddr, err)
	}
	return nil
}

// EscrowConfig configures PBS recovery-code escrow creation (slice 7, doc 03 §8a). Enrollment-time
// only (not the steady-state daemon). The default posture is zero-knowledge (Felhom holds the
// opaque blob, the customer holds the recovery code).
type EscrowConfig struct {
	// Posture is the key-custody posture; "" → zero_knowledge (the only one implemented this slice).
	Posture string `json:"posture"`
	// PBSStorageID is the pbs storage whose client encryption key is escrowed (e.g. "felhom-pbs").
	PBSStorageID string `json:"pbs_storage_id"`
}

// BackupConfig tunes the slice-6 backup + self-restore-test layer. The restore-test runs on
// an agent-internal cadence (no hub policy needed — it's self-validation); the backup
// schedule/retention/target-selection policy is hub-manifest-owned and unfed until slice 10.
type BackupConfig struct {
	// LocalBackupTarget is the vzdump storage (content=backup) backups go to, e.g. "local".
	LocalBackupTarget string `json:"local_backup_target"`
	// RestoreStorage is where a restore-test's restored rootfs lands, e.g. "local-lvm".
	RestoreStorage string `json:"restore_storage"`
	// RestoreTestCadenceSeconds is the self-restore-test interval; 0 → default (24h).
	// Set negative to DISABLE the automatic cadence (on-demand selftest still works).
	RestoreTestCadenceSeconds int `json:"restore_test_cadence_seconds"`
	// ScratchVMIDMin/Max bound the throwaway restore-test scratch-guest VMID band. The
	// restore-test refuses to run unless this is a valid band (min>0, max>=min); 9999 is
	// always excluded. Defaults to 990000–990009.
	ScratchVMIDMin int `json:"scratch_vmid_min"`
	ScratchVMIDMax int `json:"scratch_vmid_max"`

	// PBS (slice 6 Phase B). The verify maintenance loop runs on its own cadence (cheaper +
	// more frequent than the full restore-test); 0 → default (6h), negative → disabled.
	PBSVerifyCadenceSeconds int `json:"pbs_verify_cadence_seconds"`
	// PBSSecretDir holds the per-storage PBS token secret files (<id>.pw). Default
	// /etc/pve/priv/storage (PVE-managed, 0600). The agent reads it at runtime; never logged.
	PBSSecretDir string `json:"pbs_secret_dir"`
}

// Default scratch VMID band + restore-test cadence.
const (
	defaultScratchVMIDMin     = 990000
	defaultScratchVMIDMax     = 990009
	defaultRestoreTestCadence = 24 * time.Hour
)

// RestoreTestCadence returns the configured restore-test interval: a positive value as-is,
// 0 → 24h default, negative → 0 (disabled).
func (b BackupConfig) RestoreTestCadence() time.Duration {
	switch {
	case b.RestoreTestCadenceSeconds > 0:
		return time.Duration(b.RestoreTestCadenceSeconds) * time.Second
	case b.RestoreTestCadenceSeconds < 0:
		return 0 // disabled
	default:
		return defaultRestoreTestCadence
	}
}

// PBSVerifyCadence returns the verify-loop interval: positive as-is, 0 → 6h default,
// negative → 0 (disabled).
func (b BackupConfig) PBSVerifyCadence() time.Duration {
	switch {
	case b.PBSVerifyCadenceSeconds > 0:
		return time.Duration(b.PBSVerifyCadenceSeconds) * time.Second
	case b.PBSVerifyCadenceSeconds < 0:
		return -1 // disabled (pbs.VerifyLoop treats <0 as disabled)
	default:
		return 6 * time.Hour
	}
}

// PBSSecretPath returns the path to a pbs storage's token-secret file.
func (b BackupConfig) PBSSecretPath(storageID string) string {
	return b.pbsSecretDir() + "/" + storageID + ".pw"
}

// PBSEncKeyPath returns the path to a pbs storage's CLIENT ENCRYPTION KEY (K) file — the key the
// escrow wraps (slice 7). PVE stores it alongside the token secret as <id>.enc, 0600 root.
func (b BackupConfig) PBSEncKeyPath(storageID string) string {
	return b.pbsSecretDir() + "/" + storageID + ".enc"
}

func (b BackupConfig) pbsSecretDir() string {
	if b.PBSSecretDir == "" {
		return "/etc/pve/priv/storage"
	}
	return b.PBSSecretDir
}

// ScratchBand returns the effective [min,max] scratch VMID band (defaults applied).
func (b BackupConfig) ScratchBand() (min, max int) {
	min, max = b.ScratchVMIDMin, b.ScratchVMIDMax
	if min == 0 && max == 0 {
		return defaultScratchVMIDMin, defaultScratchVMIDMax
	}
	return min, max
}

// ValidateForRestoreTest checks the scratch band is usable. Called only when the restore-test
// cadence is enabled (so a host that never restore-tests needn't configure a band).
func (b BackupConfig) ValidateForRestoreTest() error {
	min, max := b.ScratchBand()
	if min <= 0 || max < min {
		return fmt.Errorf("config: backup.scratch_vmid_[min,max] is an invalid band [%d,%d]", min, max)
	}
	if 9999 >= min && 9999 <= max {
		return fmt.Errorf("config: backup scratch band [%d,%d] must not include the standing scratch 9999", min, max)
	}
	if b.RestoreStorage == "" {
		return fmt.Errorf("config: backup.restore_storage is required when the restore-test cadence is enabled")
	}
	return nil
}

// StorageConfig tunes the storage watchdog (slice 5). All optional — zero values fall back
// to the storage package defaults via the accessor methods. The watchdog poll is FAST
// (seconds) to catch a USB drop quickly; the debounce keeps a flapping drive from storming
// the hub; the known-set refresh bounds how often the watchdog re-derives the target set
// from the Proxmox API (liveness is probed every poll regardless).
type StorageConfig struct {
	WatchdogIntervalSeconds int `json:"watchdog_interval_seconds"`
	WatchdogDebounceSeconds int `json:"watchdog_debounce_seconds"`
	KnownRefreshSeconds     int `json:"known_refresh_seconds"`
}

// WatchdogInterval returns the configured poll interval (0 = package default).
func (s StorageConfig) WatchdogInterval() time.Duration {
	return time.Duration(s.WatchdogIntervalSeconds) * time.Second
}

// WatchdogDebounce returns the configured debounce window (0 = package default).
func (s StorageConfig) WatchdogDebounce() time.Duration {
	return time.Duration(s.WatchdogDebounceSeconds) * time.Second
}

// KnownRefresh returns the configured known-set refresh TTL (0 = package default).
func (s StorageConfig) KnownRefresh() time.Duration {
	return time.Duration(s.KnownRefreshSeconds) * time.Second
}

// HubConfig configures the outbound hub client + daemon poll loop (internal/hub).
// The hub serves a real cert (hub.felhom.eu, cert-manager) — this is standard TLS
// (system roots), NOT the Proxmox fingerprint-pinning path.
type HubConfig struct {
	URL            string `json:"url"`             // e.g. "https://hub.felhom.eu"
	HostID         string `json:"host_id"`         // the hub's PK for this host
	APIKey         string `json:"api_key"`         // per-host hub key; SECRET — redacted
	PollSeconds    int    `json:"poll_seconds"`    // default 900; hub may override per-cycle
	TimeoutSeconds int    `json:"timeout_seconds"` // per-request HTTP timeout; default 30
	CAFile         string `json:"ca_file"`         // optional; "" = system roots
}

// AuthzConfig configures operator-signed-op verification (internal/authz). The
// pinned operator public keys are kept here as raw authorized_keys-style lines
// (this package stays dependency-free); the authz package parses them into its
// AllowedSigner set. Role-scoping (recovery keys authorize only key-rotation) is
// enforced by the consuming layer, not loaded here.
type AuthzConfig struct {
	// NonceStorePath is the durable, crash-safe nonce log (anti-replay). Must be on
	// persistent host storage so replay protection survives agent restarts.
	NonceStorePath string `json:"nonce_store_path"`
	// Signers are the pinned operator public keys (doc 04 §3 two-key model).
	Signers []SignerKey `json:"signers"`
}

// SignerKey is one pinned operator public key.
type SignerKey struct {
	KeyID string `json:"key_id"`
	// Role is "operational" (signs destructive ops) or "recovery" (cold key;
	// authorizes only key-rotation/break-glass).
	Role string `json:"role"`
	// PublicKey is a standard authorized_keys line, e.g.
	// "ssh-ed25519 AAAA… felhom-op-1" or "sk-ssh-ed25519@openssh.com AAAA… …".
	PublicKey string `json:"public_key"`
}

// ProxmoxConfig configures the API client.
type ProxmoxConfig struct {
	// Endpoint defaults to https://127.0.0.1:8006 (agent runs on the host).
	Endpoint string `json:"endpoint"`
	// Node is the Proxmox node name; confirm on the box (GET /nodes).
	Node string `json:"node"`
	// Token is the full API token "USER@REALM!TOKENID=SECRET".
	//
	// Provisioning note: this is a privilege-SEPARATED token. Its role
	// (FelhomAgent, 16 privileges) must be granted on BOTH the user AND the token
	// for the same path, or the intersection is empty and every call 403s
	// (phase1-2 §1.2). Role setup is out-of-band; the agent only consumes the token.
	Token string `json:"token"`
	// TLS trust to the host's (self-signed) cert.
	TLS TLSTrust `json:"tls"`
}

// TLSTrust mirrors proxmox.TLSConfig (kept dependency-free here).
type TLSTrust struct {
	CAFile             string `json:"ca_file"`
	Fingerprint        string `json:"fingerprint"`          // SHA-256 of the host leaf cert
	InsecureSkipVerify bool   `json:"insecure_skip_verify"` // off by default; selftest-only
}

// PrivilegedConfig configures the fenced root-CLI runner and the slice-5 HostOps surface
// (systemd mount units + smartctl + lvs). The binary paths must match the sudoers allowlist
// exactly (see configs/felhom-agent.sudoers).
type PrivilegedConfig struct {
	// Mode: "sudo" (default — non-root agent + narrow sudoers) or "direct".
	Mode string `json:"mode"`
	// SudoPath overrides the sudo binary (default "sudo").
	SudoPath string `json:"sudo_path"`

	// HostOps (slice 5 Phase B) — the privileged storage write/read surface.
	UnitDir   string `json:"unit_dir"`  // where enabled .mount units live (default /etc/systemd/system)
	StageDir  string `json:"stage_dir"` // agent-owned staging dir for unit files (default /var/lib/felhom-agent/units)
	Systemctl string `json:"systemctl"` // default /usr/bin/systemctl
	Install   string `json:"install"`   // default /usr/bin/install
	Smartctl  string `json:"smartctl"`  // default /usr/sbin/smartctl
	Lvs       string `json:"lvs"`       // default /usr/sbin/lvs
}

// Default returns a Config pre-populated with sane defaults.
func Default() Config {
	return Config{
		Proxmox:    ProxmoxConfig{Endpoint: "https://127.0.0.1:8006"},
		Privileged: PrivilegedConfig{Mode: "sudo"},
		Authz:      AuthzConfig{NonceStorePath: "/var/lib/felhom-agent/nonces.log"},
		Hub:        HubConfig{PollSeconds: 900, TimeoutSeconds: 30},
		LogLevel:   "info",
	}
}

// Load reads the config file at path (if non-empty) over the defaults, then
// applies environment overrides. A missing path with all-env config is allowed.
func Load(path string) (Config, error) {
	cfg := Default()
	if path != "" {
		b, err := os.ReadFile(path)
		if err != nil {
			return cfg, fmt.Errorf("config: reading %s: %w", path, err)
		}
		if err := json.Unmarshal(b, &cfg); err != nil {
			return cfg, fmt.Errorf("config: parsing %s: %w", path, err)
		}
	}
	applyEnv(&cfg)
	return cfg, nil
}

// applyEnv overlays FELHOM_AGENT_* environment variables. Useful for the token in
// particular (keep the secret out of the file on disk if desired).
func applyEnv(cfg *Config) {
	if v := os.Getenv("FELHOM_AGENT_PROXMOX_ENDPOINT"); v != "" {
		cfg.Proxmox.Endpoint = v
	}
	if v := os.Getenv("FELHOM_AGENT_PROXMOX_NODE"); v != "" {
		cfg.Proxmox.Node = v
	}
	if v := os.Getenv("FELHOM_AGENT_PROXMOX_TOKEN"); v != "" {
		cfg.Proxmox.Token = v
	}
	if v := os.Getenv("FELHOM_AGENT_PROXMOX_TLS_CA_FILE"); v != "" {
		cfg.Proxmox.TLS.CAFile = v
	}
	if v := os.Getenv("FELHOM_AGENT_PROXMOX_TLS_FINGERPRINT"); v != "" {
		cfg.Proxmox.TLS.Fingerprint = v
	}
	if v := os.Getenv("FELHOM_AGENT_PROXMOX_TLS_INSECURE"); v != "" {
		if b, err := strconv.ParseBool(v); err == nil {
			cfg.Proxmox.TLS.InsecureSkipVerify = b
		}
	}
	if v := os.Getenv("FELHOM_AGENT_LOG_LEVEL"); v != "" {
		cfg.LogLevel = v
	}
	// hub
	if v := os.Getenv("FELHOM_AGENT_HUB_URL"); v != "" {
		cfg.Hub.URL = v
	}
	if v := os.Getenv("FELHOM_AGENT_HUB_HOST_ID"); v != "" {
		cfg.Hub.HostID = v
	}
	if v := os.Getenv("FELHOM_AGENT_HUB_API_KEY"); v != "" {
		cfg.Hub.APIKey = v
	}
	if v := os.Getenv("FELHOM_AGENT_HUB_CA_FILE"); v != "" {
		cfg.Hub.CAFile = v
	}
	cfg.Hub.PollSeconds = envInt("FELHOM_AGENT_HUB_POLL_SECONDS", cfg.Hub.PollSeconds)
	cfg.Hub.TimeoutSeconds = envInt("FELHOM_AGENT_HUB_TIMEOUT_SECONDS", cfg.Hub.TimeoutSeconds)
	// backup (slice 6)
	if v := os.Getenv("FELHOM_AGENT_BACKUP_LOCAL_TARGET"); v != "" {
		cfg.Backup.LocalBackupTarget = v
	}
	if v := os.Getenv("FELHOM_AGENT_BACKUP_RESTORE_STORAGE"); v != "" {
		cfg.Backup.RestoreStorage = v
	}
	cfg.Backup.RestoreTestCadenceSeconds = envInt("FELHOM_AGENT_BACKUP_RESTORE_TEST_CADENCE_SECONDS", cfg.Backup.RestoreTestCadenceSeconds)
}

// envInt overlays an int env var, keeping cur (with a stderr warning) on parse
// error rather than crashing. (Load runs before the slog logger exists.)
func envInt(key string, cur int) int {
	v := os.Getenv(key)
	if v == "" {
		return cur
	}
	n, err := strconv.Atoi(v)
	if err != nil {
		fmt.Fprintf(os.Stderr, "config: %s=%q is not an integer, keeping %d\n", key, v, cur)
		return cur
	}
	return n
}

// Validate checks the config is usable for talking to the API.
func (c Config) Validate() error {
	if c.Proxmox.Endpoint == "" {
		return fmt.Errorf("config: proxmox.endpoint is required")
	}
	if c.Proxmox.Node == "" {
		return fmt.Errorf("config: proxmox.node is required (confirm with `pvesh get /nodes`)")
	}
	if c.Proxmox.Token == "" {
		return fmt.Errorf("config: proxmox.token is required (set proxmox.token or FELHOM_AGENT_PROXMOX_TOKEN)")
	}
	if !strings.Contains(c.Proxmox.Token, "!") || !strings.Contains(c.Proxmox.Token, "=") {
		return fmt.Errorf("config: proxmox.token must be USER@REALM!TOKENID=SECRET")
	}
	return nil
}

// Redacted returns a copy safe to log: the proxmox token and hub key are masked.
func (c Config) Redacted() Config {
	if c.Proxmox.Token != "" {
		c.Proxmox.Token = redactToken(c.Proxmox.Token)
	}
	if c.Hub.APIKey != "" {
		c.Hub.APIKey = "********"
	}
	return c
}

// WithDefaults fills zero-valued hub timing fields. Applied at client/loop
// construction so programmatic configs (not from Default()) still get sane values.
func (h HubConfig) WithDefaults() HubConfig {
	if h.PollSeconds == 0 {
		h.PollSeconds = 900
	}
	if h.TimeoutSeconds == 0 {
		h.TimeoutSeconds = 30
	}
	return h
}

// Validate checks the hub config is usable for the daemon / --selftest=hub. It is
// separate from Config.Validate (proxmox-only) so --selftest=read|task still runs
// without hub config.
func (h HubConfig) Validate() error {
	if h.URL == "" {
		return fmt.Errorf("config: hub.url is required (set hub.url or FELHOM_AGENT_HUB_URL)")
	}
	if h.HostID == "" {
		return fmt.Errorf("config: hub.host_id is required")
	}
	if h.APIKey == "" {
		return fmt.Errorf("config: hub.api_key is required (set hub.api_key or FELHOM_AGENT_HUB_API_KEY)")
	}
	u, err := url.Parse(h.URL)
	if err != nil {
		return fmt.Errorf("config: hub.url is not a valid URL: %w", err)
	}
	switch u.Scheme {
	case "https":
		// always fine
	case "http":
		if !isLoopbackHost(u.Hostname()) {
			return fmt.Errorf("config: hub.url must be https:// (http:// only allowed for loopback in tests)")
		}
	default:
		return fmt.Errorf("config: hub.url must be https:// (got scheme %q)", u.Scheme)
	}
	return nil
}

func isLoopbackHost(host string) bool {
	if host == "localhost" {
		return true
	}
	if ip := net.ParseIP(host); ip != nil {
		return ip.IsLoopback()
	}
	return false
}

// redactToken keeps the public "USER@REALM!TOKENID=" prefix and masks the secret.
func redactToken(tok string) string {
	if i := strings.LastIndex(tok, "="); i >= 0 {
		return tok[:i+1] + "********"
	}
	return "********"
}