feat(provision): 弹性节点基建 Terraform + 一键更换 (tsk_6u0FxmbC7Yeq)

IaC 面 (infra/) 与控制面 (server/internal/provision) 双产出，落地 doc/04 §4 「节点是牲口」弹性拓扑与 make-before-break 一键更换。 server/internal/provision: - CloudAdapter 适配层 + Registry（首发 vultr 消耗品池 / hetzner 精品池各一）；厂商凭证仅从 PROVISION_<VENDOR>_* env 注入，不入库不入 git。 - ProvisionService：CreateNode(幂等键重放不重复开机)、DestroyNode(幂等)、 RotateIP(换 IP 不换机 + version bump)、ListProviders。 - Replace 一键更换：先建后拆，新机 up 先于旧机 draining（容量不下降）， replacement_uuid 幂等键 + replacements 表分步记录，崩溃可续跑不重复。 - RotatePool：池内滚动轮换，并发度 1–2。 - cmd/nodectl CLI：create/destroy/rotate-ip/replace/rotate-pool/providers。 - 单测（mock 厂商 API + 内存 Store）：幂等重放、make-before-break 时序断言、开机失败/探活超时→destroyed+失败计数+告警钩子、崩溃续跑、RotatePool。 infra/: - terraform/：探针机 + 控制面基线模块化（probe / control-plane）+ README，低频基线进 state，节点不进 Terraform。 - cloud-init/node.yaml.tmpl：节点引导模板（注入一次性 bootstrap token，task #5）。 - identity-isolation.md：身份隔离登记表（doc/06 §2 红线，无任何凭证）。 migrations/000008：nodes 增 provider_instance_id/elastic_ip_id、node_events 增 ip_rotated、provision_idempotency / replacements 表（附加式，不动现网）。红线：仅面向新厂商池，绝不纳管现网生产 EC2(deploy/ marzban)。 Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-13 14:23:39 +08:00
parent 787151245e
commit 3d5bac66b4
39 changed files with 3884 additions and 0 deletions
@@ -0,0 +1,366 @@
+package provision
+
+import (
+	"context"
+	"database/sql"
+	"encoding/json"
+	"fmt"
+	"strings"
+)
+
+// MySQLStore is the production Store backed by the shared *sql.DB pool.
+type MySQLStore struct {
+	db *sql.DB
+}
+
+// NewMySQLStore wraps a MySQL connection pool.
+func NewMySQLStore(db *sql.DB) *MySQLStore { return &MySQLStore{db: db} }
+
+var _ Store = (*MySQLStore)(nil)
+
+func marshalJSONList(v []string) string {
+	if v == nil {
+		v = []string{}
+	}
+	b, _ := json.Marshal(v)
+	return string(b)
+}
+
+func unmarshalJSONList(s sql.NullString) []string {
+	if !s.Valid || s.String == "" {
+		return nil
+	}
+	var out []string
+	_ = json.Unmarshal([]byte(s.String), &out)
+	return out
+}
+
+// --- nodes ---
+
+func (s *MySQLStore) InsertNode(ctx context.Context, n *Node) (int64, error) {
+	endpoint := n.Endpoint
+	if endpoint == "" {
+		endpoint = pendingEndpoint
+	}
+	status := n.Status
+	if status == "" {
+		status = StatusProvisioning
+	}
+	weight := n.Weight
+	if weight == 0 {
+		weight = 100
+	}
+	res, err := s.db.ExecContext(ctx,
+		`INSERT INTO nodes
+		  (uuid, region, name_zh, name_en, role, tier, endpoint, hy2_port,
+		   reality_pbk, reality_sni, provider_id, tags, status, weight, created_at)
+		 VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, UTC_TIMESTAMP(6))`,
+		n.UUID, n.Region, n.NameZH, n.NameEn, string(n.Role), string(n.Tier),
+		endpoint, nullInt(n.HY2Port), n.RealityPBK, n.RealitySNI, n.ProviderID,
+		marshalJSONList(n.Tags), string(status), weight)
+	if err != nil {
+		return 0, fmt.Errorf("store.InsertNode: %w", err)
+	}
+	id, err := res.LastInsertId()
+	if err != nil {
+		return 0, fmt.Errorf("store.InsertNode last id: %w", err)
+	}
+	return id, nil
+}
+
+const nodeColumns = `id, uuid, region, name_zh, name_en, role, tier, endpoint,
+	hy2_port, reality_pbk, reality_sni, provider_id, provider_instance_id,
+	elastic_ip_id, tags, status, weight, created_at`
+
+func scanNode(row interface{ Scan(...any) error }) (*Node, error) {
+	var (
+		n          Node
+		hy2        sql.NullInt64
+		instanceID sql.NullString
+		elasticID  sql.NullString
+		tags       sql.NullString
+	)
+	if err := row.Scan(
+		&n.ID, &n.UUID, &n.Region, &n.NameZH, &n.NameEn, &n.Role, &n.Tier,
+		&n.Endpoint, &hy2, &n.RealityPBK, &n.RealitySNI, &n.ProviderID,
+		&instanceID, &elasticID, &tags, &n.Status, &n.Weight, &n.CreatedAt,
+	); err != nil {
+		return nil, err
+	}
+	n.HY2Port = int(hy2.Int64)
+	n.ProviderInstanceID = instanceID.String
+	n.ElasticIPID = elasticID.String
+	n.Tags = unmarshalJSONList(tags)
+	return &n, nil
+}
+
+func (s *MySQLStore) GetNode(ctx context.Context, id int64) (*Node, error) {
+	row := s.db.QueryRowContext(ctx, `SELECT `+nodeColumns+` FROM nodes WHERE id=?`, id)
+	n, err := scanNode(row)
+	if err == sql.ErrNoRows {
+		return nil, nil
+	}
+	if err != nil {
+		return nil, fmt.Errorf("store.GetNode: %w", err)
+	}
+	return n, nil
+}
+
+func (s *MySQLStore) GetNodeByUUID(ctx context.Context, uuid string) (*Node, error) {
+	row := s.db.QueryRowContext(ctx, `SELECT `+nodeColumns+` FROM nodes WHERE uuid=?`, uuid)
+	n, err := scanNode(row)
+	if err == sql.ErrNoRows {
+		return nil, nil
+	}
+	if err != nil {
+		return nil, fmt.Errorf("store.GetNodeByUUID: %w", err)
+	}
+	return n, nil
+}
+
+func (s *MySQLStore) UpdateNodeStatus(ctx context.Context, id int64, status Status) error {
+	_, err := s.db.ExecContext(ctx, `UPDATE nodes SET status=? WHERE id=?`, string(status), id)
+	if err != nil {
+		return fmt.Errorf("store.UpdateNodeStatus: %w", err)
+	}
+	return nil
+}
+
+func (s *MySQLStore) UpdateNodeEndpoint(ctx context.Context, id int64, endpoint string) error {
+	_, err := s.db.ExecContext(ctx, `UPDATE nodes SET endpoint=? WHERE id=?`, endpoint, id)
+	if err != nil {
+		return fmt.Errorf("store.UpdateNodeEndpoint: %w", err)
+	}
+	return nil
+}
+
+func (s *MySQLStore) SetNodeInstance(ctx context.Context, id int64, instanceID, endpoint string) error {
+	_, err := s.db.ExecContext(ctx,
+		`UPDATE nodes SET provider_instance_id=?, endpoint=? WHERE id=?`,
+		instanceID, endpoint, id)
+	if err != nil {
+		return fmt.Errorf("store.SetNodeInstance: %w", err)
+	}
+	return nil
+}
+
+func (s *MySQLStore) SetNodeWeight(ctx context.Context, id int64, weight int) error {
+	_, err := s.db.ExecContext(ctx, `UPDATE nodes SET weight=? WHERE id=?`, weight, id)
+	if err != nil {
+		return fmt.Errorf("store.SetNodeWeight: %w", err)
+	}
+	return nil
+}
+
+func (s *MySQLStore) ListNodesByPool(ctx context.Context, pool Pool, status Status) ([]*Node, error) {
+	q := `SELECT ` + qualify(nodeColumns, "n") + `
+	      FROM nodes n JOIN providers p ON p.id = n.provider_id
+	      WHERE p.pool = ?`
+	args := []any{string(pool)}
+	if status != "" {
+		q += ` AND n.status = ?`
+		args = append(args, string(status))
+	}
+	q += ` ORDER BY n.id`
+	rows, err := s.db.QueryContext(ctx, q, args...)
+	if err != nil {
+		return nil, fmt.Errorf("store.ListNodesByPool: %w", err)
+	}
+	defer rows.Close()
+	var out []*Node
+	for rows.Next() {
+		n, err := scanNode(rows)
+		if err != nil {
+			return nil, fmt.Errorf("store.ListNodesByPool scan: %w", err)
+		}
+		out = append(out, n)
+	}
+	return out, rows.Err()
+}
+
+// --- providers ---
+
+func (s *MySQLStore) ListProviders(ctx context.Context, pool Pool) ([]*Provider, error) {
+	q := `SELECT id, name, api_kind, regions, pool, enabled FROM providers WHERE enabled=TRUE`
+	var args []any
+	if pool != "" {
+		q += ` AND pool=?`
+		args = append(args, string(pool))
+	}
+	q += ` ORDER BY id`
+	rows, err := s.db.QueryContext(ctx, q, args...)
+	if err != nil {
+		return nil, fmt.Errorf("store.ListProviders: %w", err)
+	}
+	defer rows.Close()
+	var out []*Provider
+	for rows.Next() {
+		p, err := scanProvider(rows)
+		if err != nil {
+			return nil, fmt.Errorf("store.ListProviders scan: %w", err)
+		}
+		out = append(out, p)
+	}
+	return out, rows.Err()
+}
+
+func (s *MySQLStore) GetProvider(ctx context.Context, id int64) (*Provider, error) {
+	row := s.db.QueryRowContext(ctx,
+		`SELECT id, name, api_kind, regions, pool, enabled FROM providers WHERE id=?`, id)
+	p, err := scanProvider(row)
+	if err == sql.ErrNoRows {
+		return nil, nil
+	}
+	if err != nil {
+		return nil, fmt.Errorf("store.GetProvider: %w", err)
+	}
+	return p, nil
+}
+
+func scanProvider(row interface{ Scan(...any) error }) (*Provider, error) {
+	var (
+		p       Provider
+		regions sql.NullString
+	)
+	if err := row.Scan(&p.ID, &p.Name, &p.APIKind, &regions, &p.Pool, &p.Enabled); err != nil {
+		return nil, err
+	}
+	p.Regions = unmarshalJSONList(regions)
+	return &p, nil
+}
+
+// --- events / audit / directory ---
+
+func (s *MySQLStore) WriteNodeEvent(ctx context.Context, nodeID int64, event Event, detailJSON string) error {
+	if detailJSON == "" {
+		detailJSON = "null"
+	}
+	_, err := s.db.ExecContext(ctx,
+		`INSERT INTO node_events (node_id, event, detail, at) VALUES (?, ?, ?, UTC_TIMESTAMP(6))`,
+		nodeID, string(event), detailJSON)
+	if err != nil {
+		return fmt.Errorf("store.WriteNodeEvent: %w", err)
+	}
+	return nil
+}
+
+func (s *MySQLStore) WriteAuditLog(ctx context.Context, actor, action, target, metaJSON string) error {
+	if metaJSON == "" {
+		metaJSON = "null"
+	}
+	_, err := s.db.ExecContext(ctx,
+		`INSERT INTO audit_log (actor, action, target, meta, at) VALUES (?, ?, ?, ?, UTC_TIMESTAMP(6))`,
+		actor, action, target, metaJSON)
+	if err != nil {
+		return fmt.Errorf("store.WriteAuditLog: %w", err)
+	}
+	return nil
+}
+
+func (s *MySQLStore) BumpDirectoryVersion(ctx context.Context) (int64, error) {
+	if _, err := s.db.ExecContext(ctx,
+		`UPDATE directory_version SET version = version + 1 WHERE id = 1`); err != nil {
+		return 0, fmt.Errorf("store.BumpDirectoryVersion: %w", err)
+	}
+	var v int64
+	if err := s.db.QueryRowContext(ctx,
+		`SELECT version FROM directory_version WHERE id = 1`).Scan(&v); err != nil {
+		return 0, fmt.Errorf("store.BumpDirectoryVersion read: %w", err)
+	}
+	return v, nil
+}
+
+// --- idempotency ---
+
+func (s *MySQLStore) LookupIdempotency(ctx context.Context, key string) (string, bool, error) {
+	var uuid string
+	err := s.db.QueryRowContext(ctx,
+		`SELECT node_uuid FROM provision_idempotency WHERE idempotency_key=?`, key).Scan(&uuid)
+	if err == sql.ErrNoRows {
+		return "", false, nil
+	}
+	if err != nil {
+		return "", false, fmt.Errorf("store.LookupIdempotency: %w", err)
+	}
+	return uuid, true, nil
+}
+
+func (s *MySQLStore) SaveIdempotency(ctx context.Context, key, nodeUUID string) error {
+	_, err := s.db.ExecContext(ctx,
+		`INSERT INTO provision_idempotency (idempotency_key, node_uuid, created_at)
+		 VALUES (?, ?, UTC_TIMESTAMP(6))
+		 ON DUPLICATE KEY UPDATE idempotency_key = idempotency_key`,
+		key, nodeUUID)
+	if err != nil {
+		return fmt.Errorf("store.SaveIdempotency: %w", err)
+	}
+	return nil
+}
+
+// --- replacements ---
+
+func (s *MySQLStore) CreateReplacement(ctx context.Context, r *Replacement) error {
+	_, err := s.db.ExecContext(ctx,
+		`INSERT INTO replacements (uuid, old_node_id, new_node_id, pool, status, step)
+		 VALUES (?, ?, ?, ?, ?, ?)`,
+		r.UUID, r.OldNodeID, nullInt64(r.NewNodeID), string(r.Pool),
+		string(r.Status), string(r.Step))
+	if err != nil {
+		return fmt.Errorf("store.CreateReplacement: %w", err)
+	}
+	return nil
+}
+
+func (s *MySQLStore) GetReplacement(ctx context.Context, uuid string) (*Replacement, error) {
+	var (
+		r     Replacement
+		newID sql.NullInt64
+	)
+	err := s.db.QueryRowContext(ctx,
+		`SELECT uuid, old_node_id, new_node_id, pool, status, step, created_at, updated_at
+		 FROM replacements WHERE uuid=?`, uuid).
+		Scan(&r.UUID, &r.OldNodeID, &newID, &r.Pool, &r.Status, &r.Step, &r.CreatedAt, &r.UpdatedAt)
+	if err == sql.ErrNoRows {
+		return nil, nil
+	}
+	if err != nil {
+		return nil, fmt.Errorf("store.GetReplacement: %w", err)
+	}
+	r.NewNodeID = newID.Int64
+	return &r, nil
+}
+
+func (s *MySQLStore) UpdateReplacement(ctx context.Context, r *Replacement) error {
+	_, err := s.db.ExecContext(ctx,
+		`UPDATE replacements SET new_node_id=?, status=?, step=? WHERE uuid=?`,
+		nullInt64(r.NewNodeID), string(r.Status), string(r.Step), r.UUID)
+	if err != nil {
+		return fmt.Errorf("store.UpdateReplacement: %w", err)
+	}
+	return nil
+}
+
+// --- helpers ---
+
+func nullInt(v int) interface{} {
+	if v == 0 {
+		return nil
+	}
+	return v
+}
+
+func nullInt64(v int64) interface{} {
+	if v == 0 {
+		return nil
+	}
+	return v
+}
+
+// qualify prefixes every comma-separated column in cols with the alias.
+func qualify(cols, alias string) string {
+	parts := strings.Split(cols, ",")
+	for i, p := range parts {
+		parts[i] = alias + "." + strings.TrimSpace(p)
+	}
+	return strings.Join(parts, ", ")
+}