Merge main into feature/demo-showcase

This commit is contained in:
Awuqing
2026-08-11 15:46:58 +08:00
218 changed files with 14685 additions and 6184 deletions
+242 -8
View File
@@ -2,15 +2,22 @@ package service
import (
"context"
"crypto/sha256"
"encoding/hex"
"encoding/json"
"errors"
"fmt"
"io"
"math"
"path"
"strings"
"time"
"backupx/server/internal/apperror"
"backupx/server/internal/backup"
"backupx/server/internal/model"
"backupx/server/internal/repository"
"backupx/server/internal/storage"
"backupx/server/internal/storage/codec"
)
@@ -23,6 +30,7 @@ type AgentService struct {
storageRepo repository.StorageTargetRepository
cmdRepo repository.AgentCommandRepository
restoreRepo repository.RestoreRecordRepository
registry *storage.Registry
cipher *codec.ConfigCipher
}
@@ -33,6 +41,7 @@ func NewAgentService(
storageRepo repository.StorageTargetRepository,
cmdRepo repository.AgentCommandRepository,
cipher *codec.ConfigCipher,
registry *storage.Registry,
) *AgentService {
return &AgentService{
nodeRepo: nodeRepo,
@@ -40,6 +49,7 @@ func NewAgentService(
recordRepo: recordRepo,
storageRepo: storageRepo,
cmdRepo: cmdRepo,
registry: registry,
cipher: cipher,
}
}
@@ -145,10 +155,11 @@ type AgentTaskSpec struct {
// AgentStorageTargetConfig 存储目标配置(已解密)
type AgentStorageTargetConfig struct {
ID uint `json:"id"`
Type string `json:"type"`
Name string `json:"name"`
Config json.RawMessage `json:"config"`
ID uint `json:"id"`
Type string `json:"type"`
Name string `json:"name"`
Config json.RawMessage `json:"config"`
TransferMode string `json:"transferMode"`
}
// GetTaskSpec 返回 Agent 执行任务所需的完整规格。
@@ -187,11 +198,22 @@ func (s *AgentService) GetTaskSpec(ctx context.Context, node *model.Node, taskID
if err != nil {
return nil, fmt.Errorf("decrypt storage config: %w", err)
}
transferMode := storage.TransferModeDirect
if strings.EqualFold(target.Type, storage.TypeLocalDisk) {
var localConfig storage.LocalDiskConfig
if err := json.Unmarshal(configRaw, &localConfig); err != nil {
return nil, fmt.Errorf("decode local disk config: %w", err)
}
if localConfig.MasterRelay {
transferMode = storage.TransferModeMasterRelay
}
}
storageTargets = append(storageTargets, AgentStorageTargetConfig{
ID: target.ID,
Type: target.Type,
Name: target.Name,
Config: json.RawMessage(configRaw),
ID: target.ID,
Type: target.Type,
Name: target.Name,
Config: json.RawMessage(configRaw),
TransferMode: transferMode,
})
}
return &AgentTaskSpec{
@@ -214,6 +236,102 @@ func (s *AgentService) GetTaskSpec(ctx context.Context, node *model.Node, taskID
}, nil
}
// UploadArtifact receives a remote Agent artifact as a stream and writes it
// with a provider created on the Master. The first supported use is local_disk,
// whose configured path belongs to the Master rather than the source Agent.
func (s *AgentService) UploadArtifact(ctx context.Context, node *model.Node, recordID, targetID uint, objectKey string, size int64, checksum string, reader io.Reader) error {
if node == nil || reader == nil || s.registry == nil {
return apperror.BadRequest("AGENT_ARTIFACT_INVALID", "中转上传参数不完整", nil)
}
record, err := s.recordRepo.FindByID(ctx, recordID)
if err != nil {
return err
}
if record == nil {
return apperror.New(404, "BACKUP_RECORD_NOT_FOUND", "记录不存在", nil)
}
task, err := s.taskRepo.FindByID(ctx, record.TaskID)
if err != nil {
return err
}
if task == nil || !recordBelongsToNode(record, task, node.ID) {
return apperror.Unauthorized("BACKUP_RECORD_FORBIDDEN", "记录不属于当前节点", nil)
}
if isBackupRecordTerminal(record.Status) {
return apperror.BadRequest("BACKUP_RECORD_TERMINAL", "备份记录已结束,不能继续上传产物", nil)
}
allowedTarget := false
for _, configuredTargetID := range collectTargetIDs(task) {
if configuredTargetID == targetID {
allowedTarget = true
break
}
}
if !allowedTarget {
return apperror.Unauthorized("BACKUP_STORAGE_TARGET_FORBIDDEN", "存储目标不属于该任务", nil)
}
target, err := s.storageRepo.FindByID(ctx, targetID)
if err != nil {
return err
}
if target == nil || !strings.EqualFold(target.Type, storage.TypeLocalDisk) {
return apperror.BadRequest("AGENT_ARTIFACT_RELAY_UNSUPPORTED", "仅 Master 本地磁盘目标需要中转上传", nil)
}
configMap := map[string]any{}
if err := s.cipher.DecryptJSON(target.ConfigCiphertext, &configMap); err != nil {
return fmt.Errorf("decrypt storage config: %w", err)
}
masterRelay, _ := configMap["masterRelay"].(bool)
if !masterRelay {
return apperror.BadRequest("AGENT_ARTIFACT_RELAY_UNSUPPORTED", "该本地磁盘目标配置为 Agent 直接写入", nil)
}
cleanKey, keyErr := s.validateArtifactKey(record, task, objectKey, true)
if keyErr != nil {
return keyErr
}
checksum = strings.TrimSpace(checksum)
checksumBytes, checksumErr := hex.DecodeString(checksum)
if size < 0 || size == math.MaxInt64 || checksumErr != nil || len(checksumBytes) != sha256.Size {
return apperror.BadRequest("AGENT_ARTIFACT_INVALID", "中转上传需要有效的大小和 SHA-256", checksumErr)
}
if target.QuotaBytes > 0 {
usage, usageErr := s.recordRepo.StorageUsage(ctx)
if usageErr != nil {
return fmt.Errorf("read storage usage: %w", usageErr)
}
currentUsed := int64(0)
for _, item := range usage {
if item.StorageTargetID == targetID {
currentUsed = item.TotalSize
break
}
}
if currentUsed >= target.QuotaBytes || size > target.QuotaBytes-currentUsed {
return apperror.BadRequest("BACKUP_STORAGE_QUOTA_EXCEEDED", fmt.Sprintf("超出存储目标配额(当前 %d,新增 %d,配额 %d)", currentUsed, size, target.QuotaBytes), nil)
}
}
provider, err := s.registry.Create(ctx, target.Type, configMap)
if err != nil {
return fmt.Errorf("create master relay provider: %w", err)
}
limited := io.LimitReader(reader, size+1)
hashed := newHashingReader(limited)
metadata := map[string]string{
"taskId": fmt.Sprintf("%d", task.ID),
"recordId": fmt.Sprintf("%d", record.ID),
"sourceNodeId": fmt.Sprintf("%d", node.ID),
"transferMode": storage.TransferModeMasterRelay,
}
if err := provider.Upload(ctx, cleanKey, hashed, size, metadata); err != nil {
return errors.Join(fmt.Errorf("relay artifact to master storage: %w", err), provider.Delete(ctx, cleanKey))
}
if hashed.n != size || !strings.EqualFold(hashed.Sum(), checksum) {
deleteErr := provider.Delete(ctx, cleanKey)
return errors.Join(fmt.Errorf("relayed artifact integrity mismatch: received %d of %d bytes", hashed.n, size), deleteErr)
}
return nil
}
func (s *AgentService) ensureTaskSpecAccess(ctx context.Context, node *model.Node, task *model.BackupTask) error {
if task.NodeID == node.ID {
return nil
@@ -236,6 +354,7 @@ type AgentRecordUpdate struct {
Checksum string `json:"checksum,omitempty"`
StoragePath string `json:"storagePath,omitempty"`
StorageTargetID uint `json:"storageTargetId,omitempty"`
StorageTransferMode string `json:"storageTransferMode,omitempty"`
StorageUploadResults []StorageUploadResultItem `json:"storageUploadResults,omitempty"`
ErrorMessage string `json:"errorMessage,omitempty"`
LogAppend string `json:"logAppend,omitempty"` // 增量日志,追加到 record.log_content
@@ -260,6 +379,99 @@ func (s *AgentService) UpdateRecord(ctx context.Context, node *model.Node, recor
if isBackupRecordTerminal(record.Status) {
return nil
}
allowedTargets := make(map[uint]struct{})
for _, targetID := range collectTargetIDs(task) {
allowedTargets[targetID] = struct{}{}
}
targetCache := make(map[uint]*model.StorageTarget)
validateTransferMode := func(targetID uint, transferMode string) (string, error) {
if _, ok := allowedTargets[targetID]; !ok {
return "", apperror.Unauthorized("BACKUP_STORAGE_TARGET_FORBIDDEN", "存储目标不属于该任务", nil)
}
target := targetCache[targetID]
if target == nil {
var findErr error
target, findErr = s.storageRepo.FindByID(ctx, targetID)
if findErr != nil {
return "", findErr
}
if target == nil {
return "", apperror.BadRequest("BACKUP_STORAGE_TARGET_INVALID", "存储目标不存在", nil)
}
targetCache[targetID] = target
}
expectedMode := storage.TransferModeDirect
if strings.EqualFold(target.Type, storage.TypeLocalDisk) {
var localConfig storage.LocalDiskConfig
if err := s.cipher.DecryptJSON(target.ConfigCiphertext, &localConfig); err != nil {
return "", fmt.Errorf("decrypt storage config: %w", err)
}
if localConfig.MasterRelay {
expectedMode = storage.TransferModeMasterRelay
}
}
if transferMode != "" && transferMode != expectedMode {
return "", apperror.BadRequest("AGENT_STORAGE_TRANSFER_MODE_INVALID", "Agent 上报的存储传输模式与目标配置不一致", nil)
}
return expectedMode, nil
}
selectedTransferMode := ""
if update.StorageTargetID > 0 {
if _, ok := allowedTargets[update.StorageTargetID]; !ok {
return apperror.Unauthorized("BACKUP_STORAGE_TARGET_FORBIDDEN", "存储目标不属于该任务", nil)
}
var modeErr error
selectedTransferMode, modeErr = validateTransferMode(update.StorageTargetID, update.StorageTransferMode)
if modeErr != nil {
return modeErr
}
} else if update.StorageTransferMode != "" {
return apperror.BadRequest("AGENT_STORAGE_TRANSFER_MODE_INVALID", "传输模式缺少对应的存储目标", nil)
}
for index := range update.StorageUploadResults {
result := &update.StorageUploadResults[index]
expectedMode, err := validateTransferMode(result.StorageTargetID, result.TransferMode)
if err != nil {
return err
}
if result.FileSize < 0 || (result.Status != "" && result.Status != "success" && result.Status != "failed") {
return apperror.BadRequest("AGENT_ARTIFACT_INVALID", "Agent 上报的存储结果无效", nil)
}
if result.StoragePath != "" {
normalizedPath, err := s.validateArtifactKey(record, task, result.StoragePath, expectedMode == storage.TransferModeMasterRelay)
if err != nil {
return err
}
result.StoragePath = normalizedPath
}
result.TransferMode = expectedMode
}
if update.StoragePath != "" {
if update.StorageTargetID == 0 {
return apperror.BadRequest("AGENT_ARTIFACT_INVALID_PATH", "存储路径缺少对应的存储目标", nil)
}
cleanStoragePath, err := s.validateArtifactKey(record, task, update.StoragePath, selectedTransferMode == storage.TransferModeMasterRelay)
if err != nil {
return err
}
if update.FileName != "" && path.Base(cleanStoragePath) != update.FileName {
return apperror.BadRequest("AGENT_ARTIFACT_INVALID_PATH", "Agent 上报的文件名与存储路径不一致", nil)
}
update.StoragePath = cleanStoragePath
}
if update.Status != "" && update.Status != model.BackupRecordStatusRunning && update.Status != model.BackupRecordStatusSuccess && update.Status != model.BackupRecordStatusFailed {
return apperror.BadRequest("BACKUP_RECORD_STATUS_INVALID", "Agent 上报的备份状态无效", nil)
}
if update.FileSize < 0 || (update.FileName != "" && (path.Base(update.FileName) != update.FileName || strings.Contains(update.FileName, "\\"))) {
return apperror.BadRequest("AGENT_ARTIFACT_INVALID", "Agent 上报的备份文件信息无效", nil)
}
if update.Checksum != "" {
checksumBytes, checksumErr := hex.DecodeString(strings.TrimSpace(update.Checksum))
if checksumErr != nil || len(checksumBytes) != sha256.Size {
return apperror.BadRequest("AGENT_ARTIFACT_INVALID", "Agent 上报的 SHA-256 无效", checksumErr)
}
update.Checksum = strings.ToLower(strings.TrimSpace(update.Checksum))
}
if update.Status != "" {
record.Status = update.Status
}
@@ -277,6 +489,7 @@ func (s *AgentService) UpdateRecord(ctx context.Context, node *model.Node, recor
}
if update.StorageTargetID > 0 {
record.StorageTargetID = update.StorageTargetID
record.StorageTransferMode = selectedTransferMode
}
if len(update.StorageUploadResults) > 0 {
if resultsJSON, marshalErr := json.Marshal(update.StorageUploadResults); marshalErr == nil {
@@ -312,6 +525,27 @@ func (s *AgentService) UpdateRecord(ctx context.Context, node *model.Node, recor
return nil
}
func (s *AgentService) validateArtifactKey(record *model.BackupRecord, task *model.BackupTask, objectKey string, requireRecordNamespace bool) (string, error) {
if record == nil || task == nil {
return "", apperror.BadRequest("AGENT_ARTIFACT_INVALID_PATH", "无法确认中转对象归属", nil)
}
rawKey := objectKey
cleanKey := path.Clean(rawKey)
fileName := path.Base(cleanKey)
if rawKey == "" || strings.TrimSpace(rawKey) != rawKey || cleanKey == "." || path.IsAbs(cleanKey) || strings.HasPrefix(cleanKey, "../") || cleanKey != rawKey || strings.Contains(rawKey, "\\") || fileName == "." || fileName == "/" {
return "", apperror.BadRequest("AGENT_ARTIFACT_INVALID_PATH", "中转上传对象路径不安全", nil)
}
expectedKey := backup.BuildRecordStorageKey(task.Type, record.StartedAt, record.ID, fileName)
if requireRecordNamespace {
legacyKey := backup.BuildStorageKey(task.Type, record.StartedAt, fileName)
if cleanKey != expectedKey && cleanKey != legacyKey {
return "", apperror.BadRequest("AGENT_ARTIFACT_INVALID_PATH", "中转上传对象不属于当前备份记录", nil)
}
return expectedKey, nil
}
return cleanKey, nil
}
func recordBelongsToNode(record *model.BackupRecord, task *model.BackupTask, nodeID uint) bool {
if record.NodeID != 0 {
return record.NodeID == nodeID
+182 -12
View File
@@ -1,19 +1,26 @@
package service
import (
"bytes"
"context"
"crypto/sha256"
"errors"
"fmt"
"os"
"path/filepath"
"strings"
"testing"
"time"
"backupx/server/internal/backup"
"backupx/server/internal/config"
"backupx/server/internal/database"
"backupx/server/internal/logger"
"backupx/server/internal/model"
"backupx/server/internal/repository"
"backupx/server/internal/storage"
"backupx/server/internal/storage/codec"
storageRclone "backupx/server/internal/storage/rclone"
"gorm.io/gorm"
)
@@ -42,7 +49,7 @@ func newAgentServicePoolTestHarness(t *testing.T) (*AgentService, *gorm.DB, repo
if err := nodeRepo.Create(context.Background(), other); err != nil {
t.Fatalf("create other node: %v", err)
}
targetConfig, err := cipher.EncryptJSON(map[string]any{"basePath": t.TempDir()})
targetConfig, err := cipher.EncryptJSON(map[string]any{"basePath": t.TempDir(), "masterRelay": true})
if err != nil {
t.Fatalf("EncryptJSON returned error: %v", err)
}
@@ -76,7 +83,8 @@ func newAgentServicePoolTestHarness(t *testing.T) (*AgentService, *gorm.DB, repo
if err := recordRepo.Create(context.Background(), record); err != nil {
t.Fatalf("create record: %v", err)
}
return NewAgentService(nodeRepo, taskRepo, recordRepo, storageRepo, cmdRepo, cipher), db, recordRepo, cmdRepo, owner, other
storageRegistry := storage.NewRegistry(storageRclone.NewLocalDiskFactory())
return NewAgentService(nodeRepo, taskRepo, recordRepo, storageRepo, cmdRepo, cipher, storageRegistry), db, recordRepo, cmdRepo, owner, other
}
func TestAgentServicePooledTaskUsesRecordNodeForSpecAndRecordUpdates(t *testing.T) {
@@ -90,19 +98,27 @@ func TestAgentServicePooledTaskUsesRecordNodeForSpecAndRecordUpdates(t *testing.
if spec.TaskID != 1 || len(spec.StorageTargets) != 1 {
t.Fatalf("unexpected spec: %#v", spec)
}
if spec.StorageTargets[0].TransferMode != storage.TransferModeMasterRelay {
t.Fatalf("expected local disk to use Master relay, got %#v", spec.StorageTargets[0])
}
if _, err := svc.GetTaskSpec(ctx, other, 1); err == nil {
t.Fatal("expected non-owner node to be forbidden from pooled task spec")
}
record, err := records.FindByID(ctx, 1)
if err != nil {
t.Fatalf("FindByID record returned error: %v", err)
}
storagePath := backup.BuildRecordStorageKey("file", record.StartedAt, record.ID, "backup.tar.gz")
if err := svc.UpdateRecord(ctx, owner, 1, AgentRecordUpdate{
Status: model.BackupRecordStatusSuccess,
FileName: "backup.tar.gz",
FileSize: 123,
StoragePath: "tasks/1/backup.tar.gz",
StorageTargetID: 2,
Status: model.BackupRecordStatusSuccess,
FileName: "backup.tar.gz",
FileSize: 123,
StoragePath: storagePath,
StorageTargetID: 1,
StorageTransferMode: storage.TransferModeMasterRelay,
StorageUploadResults: []StorageUploadResultItem{
{StorageTargetID: 1, StorageTargetName: "first", Status: "failed", Error: "boom"},
{StorageTargetID: 2, StorageTargetName: "second", Status: "success", StoragePath: "tasks/1/backup.tar.gz", FileSize: 123},
{StorageTargetID: 1, StorageTargetName: "local", Status: "success", StoragePath: storagePath, FileSize: 123, TransferMode: storage.TransferModeMasterRelay},
},
}); err != nil {
t.Fatalf("owner UpdateRecord returned error: %v", err)
@@ -114,10 +130,13 @@ func TestAgentServicePooledTaskUsesRecordNodeForSpecAndRecordUpdates(t *testing.
if updated.Status != model.BackupRecordStatusSuccess || updated.NodeID != owner.ID {
t.Fatalf("unexpected updated record: %#v", updated)
}
if updated.StorageTargetID != 2 {
t.Fatalf("expected successful storage target id 2, got %d", updated.StorageTargetID)
if updated.StorageTargetID != 1 {
t.Fatalf("expected successful storage target id 1, got %d", updated.StorageTargetID)
}
if !strings.Contains(updated.StorageUploadResults, `"storageTargetName":"second"`) {
if updated.StorageTransferMode != storage.TransferModeMasterRelay {
t.Fatalf("expected Master relay transfer mode, got %q", updated.StorageTransferMode)
}
if !strings.Contains(updated.StorageUploadResults, `"storageTargetName":"local"`) {
t.Fatalf("expected upload results to be persisted, got %q", updated.StorageUploadResults)
}
if err := svc.UpdateRecord(ctx, other, 1, AgentRecordUpdate{LogAppend: "bad"}); err == nil {
@@ -125,6 +144,157 @@ func TestAgentServicePooledTaskUsesRecordNodeForSpecAndRecordUpdates(t *testing.
}
}
func TestAgentServiceRelaysRemoteArtifactToMasterLocalDisk(t *testing.T) {
svc, _, records, _, owner, other := newAgentServicePoolTestHarness(t)
ctx := context.Background()
payload := []byte("artifact from remote source server")
digest := sha256.Sum256(payload)
checksum := fmt.Sprintf("%x", digest[:])
record, err := records.FindByID(ctx, 1)
if err != nil {
t.Fatalf("FindByID record returned error: %v", err)
}
objectKey := backup.BuildRecordStorageKey("file", record.StartedAt, record.ID, "remote-source.tar")
if err := svc.UploadArtifact(ctx, owner, 1, 1, objectKey, int64(len(payload)), checksum, bytes.NewReader(payload)); err != nil {
t.Fatalf("UploadArtifact returned error: %v", err)
}
target, err := svc.storageRepo.FindByID(ctx, 1)
if err != nil || target == nil {
t.Fatalf("FindByID target: target=%#v err=%v", target, err)
}
config := map[string]any{}
if err := svc.cipher.DecryptJSON(target.ConfigCiphertext, &config); err != nil {
t.Fatalf("DecryptJSON target config: %v", err)
}
basePath, _ := config["basePath"].(string)
stored, err := os.ReadFile(filepath.Join(basePath, filepath.FromSlash(objectKey)))
if err != nil {
t.Fatalf("read relayed artifact: %v", err)
}
if !bytes.Equal(stored, payload) {
t.Fatalf("relayed artifact differs: got %q", stored)
}
if err := svc.UploadArtifact(ctx, other, 1, 1, objectKey, int64(len(payload)), checksum, bytes.NewReader(payload)); err == nil {
t.Fatal("expected a different node to be forbidden from relaying the artifact")
}
legacyPayload := []byte("artifact from an older Agent")
legacyDigest := sha256.Sum256(legacyPayload)
legacyKey := backup.BuildStorageKey("file", record.StartedAt, "legacy-agent.tar")
canonicalKey := backup.BuildRecordStorageKey("file", record.StartedAt, record.ID, "legacy-agent.tar")
if err := svc.UploadArtifact(ctx, owner, record.ID, target.ID, legacyKey, int64(len(legacyPayload)), fmt.Sprintf("%x", legacyDigest[:]), bytes.NewReader(legacyPayload)); err != nil {
t.Fatalf("UploadArtifact legacy key returned error: %v", err)
}
stored, err = os.ReadFile(filepath.Join(basePath, filepath.FromSlash(canonicalKey)))
if err != nil || !bytes.Equal(stored, legacyPayload) {
t.Fatalf("legacy Agent artifact was not normalized: data=%q err=%v", stored, err)
}
if _, err := os.Stat(filepath.Join(basePath, filepath.FromSlash(legacyKey))); !errors.Is(err, os.ErrNotExist) {
t.Fatalf("legacy object key should not be written directly: %v", err)
}
if err := svc.UpdateRecord(ctx, owner, record.ID, AgentRecordUpdate{
Status: model.BackupRecordStatusSuccess,
FileName: "legacy-agent.tar",
FileSize: int64(len(legacyPayload)),
Checksum: fmt.Sprintf("%x", legacyDigest[:]),
StoragePath: legacyKey,
StorageTargetID: target.ID,
StorageUploadResults: []StorageUploadResultItem{{
StorageTargetID: target.ID,
Status: "success",
StoragePath: legacyKey,
FileSize: int64(len(legacyPayload)),
}},
}); err != nil {
t.Fatalf("UpdateRecord legacy key returned error: %v", err)
}
updated, err := records.FindByID(ctx, record.ID)
if err != nil {
t.Fatalf("FindByID updated record returned error: %v", err)
}
if updated.StoragePath != canonicalKey || updated.StorageTransferMode != storage.TransferModeMasterRelay || !strings.Contains(updated.StorageUploadResults, canonicalKey) {
t.Fatalf("legacy Agent record was not normalized: %#v", updated)
}
}
func TestAgentServiceRejectsArtifactOutsideRecordNamespace(t *testing.T) {
svc, _, records, _, owner, _ := newAgentServicePoolTestHarness(t)
ctx := context.Background()
record, err := records.FindByID(ctx, 1)
if err != nil {
t.Fatalf("FindByID record returned error: %v", err)
}
target, err := svc.storageRepo.FindByID(ctx, 1)
if err != nil || target == nil {
t.Fatalf("FindByID target: target=%#v err=%v", target, err)
}
config := map[string]any{}
if err := svc.cipher.DecryptJSON(target.ConfigCiphertext, &config); err != nil {
t.Fatalf("DecryptJSON target config: %v", err)
}
basePath, _ := config["basePath"].(string)
victimKey := backup.BuildRecordStorageKey("file", record.StartedAt, record.ID+1, "victim.tar")
victimPath := filepath.Join(basePath, filepath.FromSlash(victimKey))
if err := os.MkdirAll(filepath.Dir(victimPath), 0o755); err != nil {
t.Fatalf("MkdirAll victim parent: %v", err)
}
if err := os.WriteFile(victimPath, []byte("keep me"), 0o600); err != nil {
t.Fatalf("WriteFile victim: %v", err)
}
payload := []byte("overwrite")
digest := sha256.Sum256(payload)
if err := svc.UploadArtifact(ctx, owner, record.ID, target.ID, victimKey, int64(len(payload)), fmt.Sprintf("%x", digest[:]), bytes.NewReader(payload)); err == nil {
t.Fatal("expected another record namespace to be rejected")
}
stored, err := os.ReadFile(victimPath)
if err != nil {
t.Fatalf("ReadFile victim: %v", err)
}
if string(stored) != "keep me" {
t.Fatalf("victim object changed: %q", stored)
}
if err := svc.UpdateRecord(ctx, owner, record.ID, AgentRecordUpdate{StoragePath: victimKey, StorageTargetID: target.ID, StorageTransferMode: storage.TransferModeMasterRelay}); err == nil {
t.Fatal("expected another record namespace in status update to be rejected")
}
}
func TestAgentServiceKeepsExistingLocalDiskTargetsAgentLocal(t *testing.T) {
svc, _, records, _, owner, _ := newAgentServicePoolTestHarness(t)
ctx := context.Background()
target, err := svc.storageRepo.FindByID(ctx, 1)
if err != nil || target == nil {
t.Fatalf("FindByID target: target=%#v err=%v", target, err)
}
legacyConfig, err := svc.cipher.EncryptJSON(map[string]any{"basePath": t.TempDir()})
if err != nil {
t.Fatalf("EncryptJSON legacy target: %v", err)
}
target.ConfigCiphertext = legacyConfig
if err := svc.storageRepo.Update(ctx, target); err != nil {
t.Fatalf("Update legacy target: %v", err)
}
spec, err := svc.GetTaskSpec(ctx, owner, 1)
if err != nil {
t.Fatalf("GetTaskSpec returned error: %v", err)
}
if len(spec.StorageTargets) != 1 || spec.StorageTargets[0].TransferMode != storage.TransferModeDirect {
t.Fatalf("expected legacy local disk to stay Agent-local, got %#v", spec.StorageTargets)
}
payload := []byte("must not be relayed")
digest := sha256.Sum256(payload)
record, findErr := records.FindByID(ctx, 1)
if findErr != nil {
t.Fatalf("FindByID record returned error: %v", findErr)
}
objectKey := backup.BuildRecordStorageKey("file", record.StartedAt, record.ID, "legacy.tar")
err = svc.UploadArtifact(ctx, owner, 1, 1, objectKey, int64(len(payload)), fmt.Sprintf("%x", digest[:]), bytes.NewReader(payload))
if err == nil {
t.Fatal("expected relay upload to be rejected for an Agent-local target")
}
}
func TestAgentServiceUpdateRecordRefreshesTaskSummaryOnTerminalStatus(t *testing.T) {
for _, status := range []string{model.BackupRecordStatusSuccess, model.BackupRecordStatusFailed} {
t.Run(status, func(t *testing.T) {
@@ -5,11 +5,13 @@ import (
"crypto/sha256"
"encoding/hex"
"encoding/json"
"errors"
"fmt"
"hash"
"io"
"os"
"path/filepath"
"sort"
"strings"
"sync"
"time"
@@ -49,6 +51,7 @@ type StorageUploadResultItem struct {
Status string `json:"status"`
StoragePath string `json:"storagePath,omitempty"`
FileSize int64 `json:"fileSize,omitempty"`
TransferMode string `json:"transferMode,omitempty"`
Error string `json:"error,omitempty"`
}
@@ -62,6 +65,17 @@ type DownloadedArtifact struct {
Reader io.ReadCloser
}
type temporaryArtifactReader struct {
*os.File
directory string
}
func (r *temporaryArtifactReader) Close() error {
closeErr := r.File.Close()
removeErr := os.RemoveAll(r.directory)
return errors.Join(closeErr, removeErr)
}
// collectTargetIDs 获取任务关联的所有存储目标 ID
func collectTargetIDs(task *model.BackupTask) []uint {
if len(task.StorageTargets) > 0 {
@@ -102,6 +116,10 @@ type BackupExecutionService struct {
bandwidthLimit string // rclone 带宽限制(全局默认,节点配置可覆盖)
metrics *metrics.Metrics
taskLocks sync.Map
// repositoryLocks serializes immutable index updates per storage target.
// Repository mode is intentionally single-writer in v1 to avoid orphaned
// duplicate packs when two local tasks discover the same missing chunk.
repositoryLocks sync.Map
}
// SetMetrics 注入 Prometheus 采集器。nil 时所有埋点退化为 no-op。
@@ -211,6 +229,25 @@ func (s *BackupExecutionService) DownloadRecord(ctx context.Context, recordID ui
if err != nil {
return nil, err
}
if record.BackupKind == model.BackupKindRepository {
tempDir, err := os.MkdirTemp(s.tempDir, "repository-download-*")
if err != nil {
return nil, apperror.Internal("BACKUP_RECORD_DOWNLOAD_FAILED", "无法创建 CDC 导出目录", err)
}
exportName := fmt.Sprintf("backupx-record-%d.tar", record.ID)
exportPath := filepath.Join(tempDir, exportName)
store := backup.NewRepositoryStore(s.cipher.Key())
if err := store.ExportTar(ctx, provider, record.StoragePath, record.Checksum, exportPath); err != nil {
cleanupErr := os.RemoveAll(tempDir)
return nil, apperror.Internal("BACKUP_RECORD_DOWNLOAD_FAILED", "无法从 CDC 仓库导出归档", errors.Join(err, cleanupErr))
}
file, err := os.Open(exportPath)
if err != nil {
cleanupErr := os.RemoveAll(tempDir)
return nil, apperror.Internal("BACKUP_RECORD_DOWNLOAD_FAILED", "无法打开 CDC 导出归档", errors.Join(err, cleanupErr))
}
return &DownloadedArtifact{FileName: exportName, Reader: &temporaryArtifactReader{File: file, directory: tempDir}}, nil
}
reader, err := provider.Download(ctx, record.StoragePath)
if err != nil {
return nil, apperror.Internal("BACKUP_RECORD_DOWNLOAD_FAILED", "无法下载备份文件", err)
@@ -234,6 +271,16 @@ func (s *BackupExecutionService) RestoreRecord(ctx context.Context, recordID uin
if task == nil {
return apperror.New(404, "BACKUP_TASK_NOT_FOUND", "关联的备份任务不存在,无法执行恢复", fmt.Errorf("backup task %d not found", record.TaskID))
}
if record.BackupKind == model.BackupKindRepository {
spec, specErr := s.buildTaskSpec(task, record.StartedAt)
if specErr != nil {
return specErr
}
if err := backup.NewRepositoryStore(s.cipher.Key()).Restore(ctx, provider, record.StoragePath, record.Checksum, spec, backup.NopLogWriter{}); err != nil {
return apperror.Internal("BACKUP_RECORD_RESTORE_FAILED", "从 CDC 仓库恢复备份失败", err)
}
return nil
}
tempDir, err := os.MkdirTemp("", "backupx-restore-*")
if err != nil {
return apperror.Internal("BACKUP_RECORD_RESTORE_FAILED", "无法创建恢复目录", err)
@@ -291,6 +338,58 @@ func (s *BackupExecutionService) DeleteRecord(ctx context.Context, recordID uint
fmt.Sprintf("该全量备份仍有 %d 个差异备份依赖它,删除会导致这些差异无法恢复。请先删除相关差异备份或等待其过期。", deps), nil)
}
}
if record.BackupKind == model.BackupKindRepository {
copies := []StorageUploadResultItem{{
StorageTargetID: record.StorageTargetID,
Status: model.BackupRecordStatusSuccess,
StoragePath: record.StoragePath,
}}
if strings.TrimSpace(record.StorageUploadResults) != "" {
if err := json.Unmarshal([]byte(record.StorageUploadResults), &copies); err != nil {
return apperror.Internal("BACKUP_RECORD_DELETE_FAILED", "无法解析 CDC 仓库副本信息,已停止删除以避免遗留数据", err)
}
}
copyPaths := make(map[uint]string, len(copies))
for _, copy := range copies {
if strings.EqualFold(copy.Status, model.BackupRecordStatusSuccess) && strings.TrimSpace(copy.StoragePath) != "" {
copyPaths[copy.StorageTargetID] = copy.StoragePath
}
}
targetIDs := make([]uint, 0, len(copyPaths))
for targetID := range copyPaths {
targetIDs = append(targetIDs, targetID)
}
sort.Slice(targetIDs, func(i, j int) bool { return targetIDs[i] < targetIDs[j] })
unlocks := make([]func(), 0, len(targetIDs))
for _, targetID := range targetIDs {
unlocks = append(unlocks, s.acquireRepositoryLock(targetID))
}
defer func() {
for index := len(unlocks) - 1; index >= 0; index-- {
unlocks[index]()
}
}()
providers := make(map[uint]storage.StorageProvider, len(targetIDs))
for _, targetID := range targetIDs {
provider, resolveErr := s.resolveProvider(ctx, targetID)
if resolveErr != nil {
return resolveErr
}
if deleteErr := provider.Delete(ctx, copyPaths[targetID]); deleteErr != nil {
return apperror.Internal("BACKUP_RECORD_DELETE_FAILED", "无法删除 CDC 仓库快照", deleteErr)
}
providers[targetID] = provider
}
for targetID, provider := range providers {
if _, pruneErr := backup.NewRepositoryStore(s.cipher.Key()).Prune(ctx, provider); pruneErr != nil {
return apperror.Internal("BACKUP_REPOSITORY_PRUNE_FAILED", fmt.Sprintf("无法清理存储目标 %d 的 CDC 仓库;记录暂时保留以便重试", targetID), pruneErr)
}
}
if err := s.records.Delete(ctx, recordID); err != nil {
return apperror.Internal("BACKUP_RECORD_DELETE_FAILED", "无法删除备份记录", err)
}
return nil
}
if remote, err := s.deleteRemoteLocalDiskObject(ctx, record); err != nil {
return err
} else if !remote && strings.TrimSpace(record.StoragePath) != "" {
@@ -312,6 +411,9 @@ func (s *BackupExecutionService) deleteRemoteLocalDiskObject(ctx context.Context
if strings.TrimSpace(record.StoragePath) == "" || s.nodeRepo == nil {
return false, nil
}
if record.StorageTransferMode == storage.TransferModeMasterRelay {
return false, nil
}
node, err := s.nodeRepo.FindByID(ctx, record.NodeID)
if err != nil || node == nil || node.IsLocal {
return false, nil
@@ -384,6 +486,9 @@ func (s *BackupExecutionService) startTask(ctx context.Context, id uint, async b
return nil, perr
}
}
if strings.EqualFold(task.BackupMode, model.BackupModeRepository) && s.resolveRemoteNode(ctx, resolvedNodeID) != nil {
return nil, apperror.BadRequest("BACKUP_TASK_REPOSITORY_REMOTE_UNSUPPORTED", "CDC 仓库模式当前仅支持 Master 本机单写者执行", nil)
}
startedAt := s.now()
// 取第一个存储目标 ID 做兼容
primaryTargetID := task.StorageTargetID
@@ -630,6 +735,141 @@ func (s *BackupExecutionService) resolveDifferentialBase(ctx context.Context, ta
return 0, backup.Manifest{}, false
}
type repositoryTaskResult struct {
fileName string
logicalSize int64
checksum string
storagePath string
storageTargetID uint
manifestJSON string
uploadResults []StorageUploadResultItem
providers map[uint]storage.StorageProvider
}
func (s *BackupExecutionService) executeRepositoryTask(ctx context.Context, task *model.BackupTask, recordID uint, startedAt time.Time, spec backup.TaskSpec, logger *backup.ExecutionLogger) (*repositoryTaskResult, error) {
store := backup.NewRepositoryStore(s.cipher.Key())
plan, err := store.BuildPlan(ctx, spec, logger)
if err != nil {
return nil, err
}
defer func() {
if closeErr := plan.Close(); closeErr != nil {
logger.Warnf("清理 CDC 临时计划失败:%v", closeErr)
}
}()
manifestBytes, err := backup.EncodeManifest(plan.Manifest)
if err != nil {
return nil, fmt.Errorf("encode repository manifest: %w", err)
}
targetIDs := collectTargetIDs(task)
if len(targetIDs) == 0 {
return nil, fmt.Errorf("没有关联的存储目标")
}
storageUsage, usageErr := s.storageUsageSnapshot(ctx)
if usageErr != nil {
logger.Warnf("读取存储目标用量失败,跳过本次软配额校验:%v", usageErr)
storageUsage = map[uint]int64{}
}
snapshotKey := store.SnapshotKey(task.ID, recordID, startedAt)
result := &repositoryTaskResult{
fileName: filepath.Base(snapshotKey),
logicalSize: plan.LogicalSize,
storagePath: snapshotKey,
manifestJSON: string(manifestBytes),
uploadResults: make([]StorageUploadResultItem, 0, len(targetIDs)),
providers: make(map[uint]storage.StorageProvider),
}
var failures []string
for _, targetID := range targetIDs {
target, findErr := s.targets.FindByID(ctx, targetID)
targetName := fmt.Sprintf("target-%d", targetID)
if findErr == nil && target != nil {
targetName = target.Name
}
if findErr != nil || target == nil {
message := "存储目标不存在"
if findErr != nil {
message = findErr.Error()
}
result.uploadResults = append(result.uploadResults, StorageUploadResultItem{StorageTargetID: targetID, StorageTargetName: targetName, Status: "failed", Error: message})
failures = append(failures, fmt.Sprintf("%s: %s", targetName, message))
continue
}
provider, resolveErr := s.resolveProviderForNode(ctx, targetID, task.NodeID)
if resolveErr != nil {
result.uploadResults = append(result.uploadResults, StorageUploadResultItem{StorageTargetID: targetID, StorageTargetName: targetName, Status: "failed", Error: resolveErr.Error()})
failures = append(failures, fmt.Sprintf("%s: %v", targetName, resolveErr))
continue
}
logger.Infof("同步 CDC 仓库到存储目标:%s", targetName)
unlock := s.acquireRepositoryLock(targetID)
estimatedSize, estimateErr := store.EstimateUploadSize(ctx, provider, plan)
if estimateErr != nil {
unlock()
result.uploadResults = append(result.uploadResults, StorageUploadResultItem{StorageTargetID: targetID, StorageTargetName: targetName, Status: "failed", Error: estimateErr.Error()})
failures = append(failures, fmt.Sprintf("%s: %v", targetName, estimateErr))
continue
}
if target.QuotaBytes > 0 && storageUsage[targetID]+estimatedSize > target.QuotaBytes {
unlock()
message := fmt.Sprintf("超出存储目标配额(%d + 预计 %d > %d", storageUsage[targetID], estimatedSize, target.QuotaBytes)
result.uploadResults = append(result.uploadResults, StorageUploadResultItem{StorageTargetID: targetID, StorageTargetName: targetName, Status: "failed", Error: message})
failures = append(failures, fmt.Sprintf("%s: %s", targetName, message))
continue
}
upload, uploadErr := store.Upload(ctx, provider, plan, snapshotKey)
unlock()
if uploadErr != nil {
result.uploadResults = append(result.uploadResults, StorageUploadResultItem{StorageTargetID: targetID, StorageTargetName: targetName, Status: "failed", Error: uploadErr.Error()})
failures = append(failures, fmt.Sprintf("%s: %v", targetName, uploadErr))
logger.Warnf("存储目标 %s CDC 仓库同步失败:%v", targetName, uploadErr)
continue
}
result.uploadResults = append(result.uploadResults, StorageUploadResultItem{
StorageTargetID: targetID, StorageTargetName: targetName, Status: "success",
StoragePath: upload.SnapshotKey, FileSize: upload.UploadedBytes,
})
result.providers[targetID] = provider
if result.storageTargetID == 0 {
result.storageTargetID = targetID
result.checksum = upload.Checksum
}
logger.Infof("存储目标 %s CDC 同步完成:新块 %d/%d,复用 %d bytes,实际上传 %d bytes", targetName, upload.NewChunks, upload.UniqueChunks, upload.ReusedBytes, upload.UploadedBytes)
}
if result.storageTargetID == 0 {
return nil, fmt.Errorf("所有存储目标 CDC 仓库同步均失败:%s", strings.Join(failures, "; "))
}
if len(failures) > 0 {
logger.Warnf("部分存储目标 CDC 仓库同步失败:%s", strings.Join(failures, "; "))
}
if s.dependentsResolver != nil {
go func(upstreamID uint, upstreamName string) {
dependents, resolveErr := s.dependentsResolver.TriggerDependents(context.Background(), upstreamID)
if resolveErr != nil {
logger.Warnf("解析任务 %s 的下游依赖失败:%v", upstreamName, resolveErr)
return
}
for _, dependentID := range dependents {
if _, runErr := s.RunTaskByID(context.Background(), dependentID); runErr != nil {
logger.Warnf("触发下游任务 #%d 失败(上游: %s):%v", dependentID, upstreamName, runErr)
} else {
logger.Infof("已触发下游任务 #%d(上游: %s", dependentID, upstreamName)
}
}
}(task.ID, task.Name)
}
return result, nil
}
func (s *BackupExecutionService) acquireRepositoryLock(targetID uint) func() {
created := &sync.Mutex{}
actual, _ := s.repositoryLocks.LoadOrStore(targetID, created)
lock := actual.(*sync.Mutex)
lock.Lock()
return lock.Unlock
}
func (s *BackupExecutionService) executeTask(ctx context.Context, task *model.BackupTask, recordID uint, startedAt time.Time) {
// 节点级并发限流:当任务绑定节点且节点配置了 MaxConcurrent>0
// 该节点上所有任务共享一个节点专属 semaphore,互相排队
@@ -658,18 +898,33 @@ func (s *BackupExecutionService) executeTask(ctx context.Context, task *model.Ba
backupKind := model.BackupKindFull
var baseRecordID uint
var manifestJSON string
var repositoryProviders map[uint]storage.StorageProvider
completeRecord := func() {
readyForRepositoryRetention := status == model.BackupRecordStatusSuccess
if finalizeErr := s.finalizeRecord(ctx, task, recordID, startedAt, status, errMessage, logger.String(), fileName, fileSize, checksum, storagePath, selectedStorageTargetID); finalizeErr != nil {
logger.Errorf("写回备份记录失败:%v", finalizeErr)
readyForRepositoryRetention = false
}
// 采集任务执行结果到 Prometheus(耗时 + 产出字节 + 状态计数)
s.metrics.ObserveTaskRun(task.Type, status, time.Since(startedAt).Seconds(), fileSize)
// 写入多目标上传结果
if len(uploadResults) > 0 {
if resultsJSON, marshalErr := json.Marshal(uploadResults); marshalErr == nil {
if record, findErr := s.records.FindByID(ctx, recordID); findErr == nil && record != nil {
record.StorageUploadResults = string(resultsJSON)
_ = s.records.Update(ctx, record)
resultsJSON, marshalErr := json.Marshal(uploadResults)
if marshalErr != nil {
logger.Warnf("序列化多目标上传结果失败:%v", marshalErr)
readyForRepositoryRetention = false
} else if record, findErr := s.records.FindByID(ctx, recordID); findErr != nil || record == nil {
if findErr != nil {
logger.Warnf("读取备份记录以写回多目标结果失败:%v", findErr)
} else {
logger.Warnf("备份记录 #%d 不存在,无法写回多目标结果", recordID)
}
readyForRepositoryRetention = false
} else {
record.StorageUploadResults = string(resultsJSON)
if updateErr := s.records.Update(ctx, record); updateErr != nil {
logger.Warnf("写回多目标上传结果失败:%v", updateErr)
readyForRepositoryRetention = false
}
}
}
@@ -681,6 +936,36 @@ func (s *BackupExecutionService) executeTask(ctx context.Context, task *model.Ba
record.Manifest = manifestJSON
if updErr := s.records.Update(ctx, record); updErr != nil {
logger.Warnf("写回差异链信息失败:%v", updErr)
readyForRepositoryRetention = false
}
} else {
if findErr != nil {
logger.Warnf("读取备份记录以写回备份类型失败:%v", findErr)
} else {
logger.Warnf("备份记录 #%d 不存在,无法写回备份类型", recordID)
}
readyForRepositoryRetention = false
}
}
if readyForRepositoryRetention && backupKind == model.BackupKindRepository && s.retention != nil && len(repositoryProviders) > 0 {
targetIDs := make([]uint, 0, len(repositoryProviders))
for targetID := range repositoryProviders {
targetIDs = append(targetIDs, targetID)
}
sort.Slice(targetIDs, func(i, j int) bool { return targetIDs[i] < targetIDs[j] })
unlocks := make([]func(), 0, len(targetIDs))
for _, targetID := range targetIDs {
unlocks = append(unlocks, s.acquireRepositoryLock(targetID))
}
cleanupResult, cleanupErr := s.retention.CleanupProviders(ctx, task, repositoryProviders)
for index := len(unlocks) - 1; index >= 0; index-- {
unlocks[index]()
}
if cleanupErr != nil {
logger.Warnf("执行 CDC 仓库保留策略失败:%v", cleanupErr)
} else {
for _, warning := range cleanupResult.Warnings {
logger.Warnf("CDC 仓库保留策略警告:%s", warning)
}
}
}
@@ -701,6 +986,26 @@ func (s *BackupExecutionService) executeTask(ctx context.Context, task *model.Ba
logger.Errorf("构建任务运行时配置失败:%v", err)
return
}
if task.Type == model.BackupTaskTypeFile && strings.EqualFold(task.BackupMode, model.BackupModeRepository) {
backupKind = model.BackupKindRepository
repositoryResult, repositoryErr := s.executeRepositoryTask(ctx, task, recordID, startedAt, spec, logger)
if repositoryErr != nil {
errMessage = repositoryErr.Error()
logger.Errorf("执行 CDC 仓库备份失败:%v", repositoryErr)
return
}
fileName = repositoryResult.fileName
fileSize = repositoryResult.logicalSize
checksum = repositoryResult.checksum
storagePath = repositoryResult.storagePath
selectedStorageTargetID = repositoryResult.storageTargetID
uploadResults = repositoryResult.uploadResults
repositoryProviders = repositoryResult.providers
manifestJSON = repositoryResult.manifestJSON
status = model.BackupRecordStatusSuccess
logger.Infof("CDC 仓库备份执行完成")
return
}
// 差异备份:解析基线全量,命中则切换为差异模式(仅本机文件任务)。
if baseID, baseManifest, ok := s.resolveDifferentialBase(ctx, task); ok {
spec.Differential = true
@@ -1,6 +1,7 @@
package service
import (
"bytes"
"context"
"errors"
"fmt"
@@ -188,6 +189,108 @@ func TestBackupExecutionServiceSQLiteBackupRemainsFull(t *testing.T) {
}
}
func TestBackupExecutionServiceRepositoryModeRoundTrip(t *testing.T) {
executionService, recordService, tasks, _, records, sourceDir, storageDir := newExecutionTestServices(t)
ctx := context.Background()
task, err := tasks.FindByID(ctx, 1)
if err != nil {
t.Fatalf("FindByID task returned error: %v", err)
}
task.BackupMode = model.BackupModeRepository
task.Compression = "zstd"
if err := tasks.Update(ctx, task); err != nil {
t.Fatalf("Update repository task returned error: %v", err)
}
large := make([]byte, 4<<20)
for index := range large {
large[index] = byte((index * 31) % 251)
}
largePath := filepath.Join(sourceDir, "large.bin")
if err := os.WriteFile(largePath, large, 0o640); err != nil {
t.Fatalf("write large fixture: %v", err)
}
first, err := executionService.RunTaskByIDSync(ctx, task.ID)
if err != nil {
t.Fatalf("first repository backup returned error: %v", err)
}
if first.Status != model.BackupRecordStatusSuccess || first.BackupKind != model.BackupKindRepository {
t.Fatalf("unexpected first repository record: %#v", first)
}
if !strings.HasPrefix(first.StoragePath, ".backupx/repository/v1/snapshots/") {
t.Fatalf("unexpected repository snapshot path: %s", first.StoragePath)
}
large[2<<20] ^= 0xff
if err := os.WriteFile(largePath, large, 0o640); err != nil {
t.Fatalf("modify large fixture: %v", err)
}
second, err := executionService.RunTaskByIDSync(ctx, task.ID)
if err != nil {
t.Fatalf("second repository backup returned error: %v", err)
}
stored, err := records.FindByID(ctx, second.ID)
if err != nil {
t.Fatalf("FindByID repository record returned error: %v", err)
}
if stored == nil || stored.BackupKind != model.BackupKindRepository || stored.Manifest == "" {
t.Fatalf("repository metadata was not persisted: %#v", stored)
}
download, err := recordService.Download(ctx, second.ID)
if err != nil {
t.Fatalf("export repository snapshot returned error: %v", err)
}
exported, readErr := io.ReadAll(download.Reader)
closeErr := download.Reader.Close()
if readErr != nil || closeErr != nil {
t.Fatalf("read repository export: read=%v close=%v", readErr, closeErr)
}
if len(exported) == 0 || !strings.HasSuffix(download.FileName, ".tar") {
t.Fatalf("unexpected repository export: name=%s size=%d", download.FileName, len(exported))
}
if err := os.WriteFile(largePath, bytes.Repeat([]byte{0}, len(large)), 0o640); err != nil {
t.Fatalf("damage source before restore: %v", err)
}
if err := executionService.RestoreRecord(ctx, second.ID); err != nil {
t.Fatalf("restore repository record returned error: %v", err)
}
restored, err := os.ReadFile(largePath)
if err != nil {
t.Fatalf("read restored source: %v", err)
}
if !bytes.Equal(restored, large) {
t.Fatalf("repository restore did not reproduce the source")
}
if err := recordService.Delete(ctx, first.ID); err != nil {
t.Fatalf("delete first repository record: %v", err)
}
if err := recordService.Delete(ctx, second.ID); err != nil {
t.Fatalf("delete second repository record: %v", err)
}
packRoot := filepath.Join(storageDir, filepath.FromSlash(".backupx/repository/v1/packs"))
remainingPacks := 0
if err := filepath.Walk(packRoot, func(_ string, info os.FileInfo, walkErr error) error {
if walkErr != nil {
if os.IsNotExist(walkErr) {
return nil
}
return walkErr
}
if info != nil && !info.IsDir() {
remainingPacks++
}
return nil
}); err != nil {
t.Fatalf("inspect repository packs: %v", err)
}
if remainingPacks != 0 {
t.Fatalf("repository prune left %d packs after deleting all snapshots", remainingPacks)
}
}
func TestBackupExecutionServiceNodePoolSelectionDoesNotPersistTaskNodeID(t *testing.T) {
executionService, _, tasks, _, records, _, _ := newExecutionTestServices(t)
ctx := context.Background()
@@ -359,6 +462,56 @@ func TestBackupExecutionServiceRestoreRecordRejectsRemoteLocalDisk(t *testing.T)
}
}
func TestBackupExecutionServiceDownloadsMasterRelayedLocalDiskRecord(t *testing.T) {
executionService, _, tasks, _, records, _, storageDir := newExecutionTestServices(t)
ctx := context.Background()
executionService.SetClusterDependencies(&nodeRepoStub{nodes: []model.Node{
{ID: 10, Name: "edge-a", Token: "edge-a-token", Status: model.NodeStatusOnline},
}}, &fakeDispatcher{})
task, err := tasks.FindByID(ctx, 1)
if err != nil {
t.Fatalf("FindByID task returned error: %v", err)
}
storagePath := "file/2026/05/09/relayed.tar"
artifactPath := filepath.Join(storageDir, filepath.FromSlash(storagePath))
if err := os.MkdirAll(filepath.Dir(artifactPath), 0o755); err != nil {
t.Fatalf("MkdirAll artifact parent returned error: %v", err)
}
content := []byte("stored on Master")
if err := os.WriteFile(artifactPath, content, 0o600); err != nil {
t.Fatalf("WriteFile artifact returned error: %v", err)
}
completedAt := time.Now().UTC()
record := &model.BackupRecord{
TaskID: task.ID,
StorageTargetID: task.StorageTargetID,
NodeID: 10,
Status: model.BackupRecordStatusSuccess,
FileName: "relayed.tar",
FileSize: int64(len(content)),
StoragePath: storagePath,
StorageTransferMode: storage.TransferModeMasterRelay,
StartedAt: completedAt.Add(-time.Second),
CompletedAt: &completedAt,
}
if err := records.Create(ctx, record); err != nil {
t.Fatalf("Create record returned error: %v", err)
}
download, err := executionService.DownloadRecord(ctx, record.ID)
if err != nil {
t.Fatalf("DownloadRecord returned error: %v", err)
}
got, readErr := io.ReadAll(download.Reader)
closeErr := download.Reader.Close()
if readErr != nil || closeErr != nil {
t.Fatalf("read relayed artifact: read=%v close=%v", readErr, closeErr)
}
if !bytes.Equal(got, content) {
t.Fatalf("downloaded content = %q, want %q", got, content)
}
}
func TestBackupExecutionServiceRecordsFirstSuccessfulStorageTarget(t *testing.T) {
executionService, _, tasks, targets, records, _, _ := newExecutionTestServices(t)
ctx := context.Background()
@@ -23,22 +23,23 @@ type BackupRecordListInput struct {
}
type BackupRecordSummary struct {
ID uint `json:"id"`
TaskID uint `json:"taskId"`
TaskName string `json:"taskName"`
StorageTargetID uint `json:"storageTargetId"`
StorageTargetName string `json:"storageTargetName"`
Status string `json:"status"`
FileName string `json:"fileName"`
FileSize int64 `json:"fileSize"`
Checksum string `json:"checksum"`
StoragePath string `json:"storagePath"`
DurationSeconds int `json:"durationSeconds"`
ErrorMessage string `json:"errorMessage"`
StartedAt time.Time `json:"startedAt"`
CompletedAt *time.Time `json:"completedAt,omitempty"`
Locked bool `json:"locked"`
BackupKind string `json:"backupKind"`
ID uint `json:"id"`
TaskID uint `json:"taskId"`
TaskName string `json:"taskName"`
StorageTargetID uint `json:"storageTargetId"`
StorageTargetName string `json:"storageTargetName"`
Status string `json:"status"`
FileName string `json:"fileName"`
FileSize int64 `json:"fileSize"`
Checksum string `json:"checksum"`
StoragePath string `json:"storagePath"`
StorageTransferMode string `json:"storageTransferMode,omitempty"`
DurationSeconds int `json:"durationSeconds"`
ErrorMessage string `json:"errorMessage"`
StartedAt time.Time `json:"startedAt"`
CompletedAt *time.Time `json:"completedAt,omitempty"`
Locked bool `json:"locked"`
BackupKind string `json:"backupKind"`
}
type BackupRecordDetail struct {
@@ -184,22 +185,23 @@ func (s *BackupRecordService) SetLock(ctx context.Context, id uint, locked bool)
func toBackupRecordSummary(item *model.BackupRecord) BackupRecordSummary {
return BackupRecordSummary{
ID: item.ID,
TaskID: item.TaskID,
TaskName: item.Task.Name,
StorageTargetID: item.StorageTargetID,
StorageTargetName: item.StorageTarget.Name,
Status: item.Status,
FileName: item.FileName,
FileSize: item.FileSize,
Checksum: item.Checksum,
StoragePath: item.StoragePath,
DurationSeconds: item.DurationSeconds,
ErrorMessage: item.ErrorMessage,
StartedAt: item.StartedAt,
CompletedAt: item.CompletedAt,
Locked: item.Locked,
BackupKind: item.BackupKind,
ID: item.ID,
TaskID: item.TaskID,
TaskName: item.Task.Name,
StorageTargetID: item.StorageTargetID,
StorageTargetName: item.StorageTarget.Name,
Status: item.Status,
FileName: item.FileName,
FileSize: item.FileSize,
Checksum: item.Checksum,
StoragePath: item.StoragePath,
StorageTransferMode: item.StorageTransferMode,
DurationSeconds: item.DurationSeconds,
ErrorMessage: item.ErrorMessage,
StartedAt: item.StartedAt,
CompletedAt: item.CompletedAt,
Locked: item.Locked,
BackupKind: item.BackupKind,
}
}
+60 -13
View File
@@ -57,8 +57,8 @@ type BackupTaskUpsertInput struct {
KeepWeekly int `json:"keepWeekly"`
KeepMonthly int `json:"keepMonthly"`
KeepYearly int `json:"keepYearly"`
// BackupMode 备份模式:full(默认)/ differential(差异,仅文件类型本机任务
BackupMode string `json:"backupMode" binding:"omitempty,oneof=full differential"`
// BackupMode 备份模式:full(默认)/ differential(差异归档)/ repositoryCDC 去重仓库
BackupMode string `json:"backupMode" binding:"omitempty,oneof=full differential repository"`
DiffFullIntervalDays int `json:"diffFullIntervalDays"`
// 备份复制目标存储 ID 列表(3-2-1 规则)
ReplicationTargetIDs []uint `json:"replicationTargetIds"`
@@ -414,21 +414,50 @@ func (s *BackupTaskService) cleanupRemoteFiles(ctx context.Context, taskID uint)
recordCount = len(records)
// 缓存 provider 避免同一存储目标重复创建连接
providerCache := make(map[uint]storage.StorageProvider)
repositoryProviders := make(map[uint]storage.StorageProvider)
for _, record := range records {
if strings.TrimSpace(record.StoragePath) == "" {
continue
copies := []StorageUploadResultItem{{
StorageTargetID: record.StorageTargetID,
Status: model.BackupRecordStatusSuccess,
StoragePath: record.StoragePath,
}}
if strings.TrimSpace(record.StorageUploadResults) != "" {
var storedCopies []StorageUploadResultItem
if unmarshalErr := json.Unmarshal([]byte(record.StorageUploadResults), &storedCopies); unmarshalErr == nil {
copies = storedCopies
}
}
provider, ok := providerCache[record.StorageTargetID]
if !ok {
provider, err = s.resolveStorageProvider(ctx, record.StorageTargetID)
if err != nil {
seenTargets := make(map[uint]struct{}, len(copies))
for _, copy := range copies {
if !strings.EqualFold(copy.Status, model.BackupRecordStatusSuccess) || strings.TrimSpace(copy.StoragePath) == "" {
continue
}
providerCache[record.StorageTargetID] = provider
if _, seen := seenTargets[copy.StorageTargetID]; seen {
continue
}
seenTargets[copy.StorageTargetID] = struct{}{}
provider, ok := providerCache[copy.StorageTargetID]
if !ok {
provider, err = s.resolveStorageProvider(ctx, copy.StorageTargetID)
if err != nil {
continue
}
providerCache[copy.StorageTargetID] = provider
}
if err := provider.Delete(ctx, copy.StoragePath); err == nil {
cleanedFiles++
if record.BackupKind == model.BackupKindRepository {
repositoryProviders[copy.StorageTargetID] = provider
}
}
}
if err := provider.Delete(ctx, record.StoragePath); err == nil {
cleanedFiles++
}
for _, provider := range repositoryProviders {
pruned, pruneErr := backup.NewRepositoryStore(s.cipher.Key()).Prune(ctx, provider)
if pruneErr != nil {
continue
}
cleanedFiles += pruned.DeletedIndexes + pruned.DeletedPacks
}
return recordCount, cleanedFiles
}
@@ -530,6 +559,17 @@ func (s *BackupTaskService) validateInput(ctx context.Context, existing *model.B
return apperror.BadRequest("BACKUP_TASK_DIFF_REMOTE_UNSUPPORTED", "差异备份当前仅支持本机 Master 执行,请将任务固定在本机或改用全量备份。", nil)
}
}
if strings.EqualFold(strings.TrimSpace(input.BackupMode), model.BackupModeRepository) {
if input.Type != model.BackupTaskTypeFile {
return apperror.BadRequest("BACKUP_TASK_REPOSITORY_UNSUPPORTED", "CDC 仓库模式仅支持文件目录类型任务", nil)
}
if strings.TrimSpace(input.NodePoolTag) != "" || (fixedNode != nil && !fixedNode.IsLocal) {
return apperror.BadRequest("BACKUP_TASK_REPOSITORY_REMOTE_UNSUPPORTED", "CDC 仓库模式当前采用单写者索引,仅支持 Master 本机执行。远程服务器备份请暂用全量模式。", nil)
}
if len(input.ReplicationTargetIDs) > 0 {
return apperror.BadRequest("BACKUP_TASK_REPOSITORY_REPLICATION_UNSUPPORTED", "CDC 仓库快照不能使用对象级复制;请直接为任务选择多个存储目标以生成完整仓库副本。", nil)
}
}
if input.RetentionDays < 0 {
return apperror.BadRequest("BACKUP_TASK_INVALID", "保留天数不能小于 0", nil)
}
@@ -935,10 +975,17 @@ func decodeExtraConfig(value string) (map[string]any, error) {
return result, nil
}
// normalizeBackupMode 归一化备份模式:仅文件类型可启用差异,其余一律全量(双保险,防绕过校验)。
// normalizeBackupMode 归一化备份模式:仅文件类型可启用差异或 CDC 仓库,
// 其余一律全量(双保险,防绕过校验)。
func normalizeBackupMode(mode, taskType string) string {
if strings.EqualFold(strings.TrimSpace(mode), model.BackupModeDifferential) && normalizeBackupTaskType(taskType) == model.BackupTaskTypeFile {
if normalizeBackupTaskType(taskType) != model.BackupTaskTypeFile {
return model.BackupModeFull
}
switch strings.ToLower(strings.TrimSpace(mode)) {
case model.BackupModeDifferential:
return model.BackupModeDifferential
case model.BackupModeRepository:
return model.BackupModeRepository
}
return model.BackupModeFull
}
+20 -20
View File
@@ -126,14 +126,14 @@ func (s *DashboardService) Timeline(ctx context.Context, days int) ([]repository
// 判定规则:任务设置了 SLAHoursRPO > 0,且距最近一次 success 备份的时间 > SLAHoursRPO。
// 从未成功过的任务(LastSuccessAt = nil)若启用也视为违约(from createdAt 起算)。
type SLAViolation struct {
TaskID uint `json:"taskId"`
TaskName string `json:"taskName"`
NodeID uint `json:"nodeId"`
NodeName string `json:"nodeName,omitempty"`
SLAHoursRPO int `json:"slaHoursRpo"`
LastSuccessAt *time.Time `json:"lastSuccessAt,omitempty"`
HoursSinceLastSuccess float64 `json:"hoursSinceLastSuccess"`
NeverSucceeded bool `json:"neverSucceeded"`
TaskID uint `json:"taskId"`
TaskName string `json:"taskName"`
NodeID uint `json:"nodeId"`
NodeName string `json:"nodeName,omitempty"`
SLAHoursRPO int `json:"slaHoursRpo"`
LastSuccessAt *time.Time `json:"lastSuccessAt,omitempty"`
HoursSinceLastSuccess float64 `json:"hoursSinceLastSuccess"`
NeverSucceeded bool `json:"neverSucceeded"`
}
// SLAComplianceReport Dashboard 的 SLA 合规概览。
@@ -204,15 +204,15 @@ func roundHours(value float64) float64 {
// ClusterNodeSummary 集群节点简报(Dashboard 用)。
type ClusterNodeSummary struct {
ID uint `json:"id"`
Name string `json:"name"`
Hostname string `json:"hostname"`
Status string `json:"status"`
IsLocal bool `json:"isLocal"`
AgentVersion string `json:"agentVersion"`
VersionStatus string `json:"versionStatus"` // current | outdated | unknown
LastSeen time.Time `json:"lastSeen"`
TaskCount int64 `json:"taskCount"`
ID uint `json:"id"`
Name string `json:"name"`
Hostname string `json:"hostname"`
Status string `json:"status"`
IsLocal bool `json:"isLocal"`
AgentVersion string `json:"agentVersion"`
VersionStatus string `json:"versionStatus"` // current | outdated | unknown
LastSeen time.Time `json:"lastSeen"`
TaskCount int64 `json:"taskCount"`
}
// ClusterOverview Dashboard 集群概览卡片。
@@ -312,9 +312,9 @@ func (s *DashboardService) Breakdown(ctx context.Context, days int) (*BreakdownS
}
}
result := &BreakdownStats{
ByType: makeBreakdown(typeCounts, typeLabel),
ByNode: makeBreakdownByUint(nodeCounts, nodeNames, "节点 #"),
ByStatus: []BreakdownItem{},
ByType: makeBreakdown(typeCounts, typeLabel),
ByNode: makeBreakdownByUint(nodeCounts, nodeNames, "节点 #"),
ByStatus: []BreakdownItem{},
ByStorage: []BreakdownItem{},
}
// 按状态(最近 days 天记录)
@@ -140,6 +140,11 @@ func validateCrossNodeLocalDisk(ctx context.Context, nodeRepo repository.NodeRep
if record == nil || record.NodeID == 0 || nodeRepo == nil {
return nil
}
// 中转模式的对象实际落在 Master 配置的本地磁盘,Master 可以安全访问。
// 空值和 direct 均按旧版 Agent 本地落盘处理,保持升级兼容。
if record.StorageTransferMode == storage.TransferModeMasterRelay {
return nil
}
node, err := nodeRepo.FindByID(ctx, record.NodeID)
if err != nil || node == nil || node.IsLocal {
return nil
@@ -6,6 +6,7 @@ import (
"encoding/base64"
"encoding/hex"
"fmt"
"net/url"
"strings"
"time"
@@ -27,13 +28,16 @@ func NewInstallTokenService(repo repository.AgentInstallTokenRepository, nodeRep
// InstallTokenInput 生成一次性安装令牌的输入。
type InstallTokenInput struct {
NodeID uint
Mode string
Arch string
AgentVersion string
DownloadSrc string
TTLSeconds int
CreatedByID uint
NodeID uint
Mode string
Arch string
AgentVersion string
DownloadSrc string
TTLSeconds int
CreatedByID uint
AgentMasterURL string
ProxyURL string
CACertFile string
}
// InstallTokenOutput 生成结果。
@@ -112,14 +116,17 @@ func (s *InstallTokenService) Create(ctx context.Context, in InstallTokenInput)
}
expiresAt := time.Now().UTC().Add(time.Duration(in.TTLSeconds) * time.Second)
record := &model.AgentInstallToken{
Token: token,
NodeID: in.NodeID,
Mode: in.Mode,
Arch: in.Arch,
AgentVer: in.AgentVersion,
DownloadSrc: in.DownloadSrc,
ExpiresAt: expiresAt,
CreatedByID: in.CreatedByID,
Token: token,
NodeID: in.NodeID,
Mode: in.Mode,
Arch: in.Arch,
AgentVer: in.AgentVersion,
DownloadSrc: in.DownloadSrc,
AgentMasterURL: strings.TrimRight(strings.TrimSpace(in.AgentMasterURL), "/"),
ProxyURL: strings.TrimSpace(in.ProxyURL),
CACertFile: strings.TrimSpace(in.CACertFile),
ExpiresAt: expiresAt,
CreatedByID: in.CreatedByID,
}
if err := s.repo.Create(ctx, record); err != nil {
return nil, err
@@ -130,9 +137,15 @@ func (s *InstallTokenService) Create(ctx context.Context, in InstallTokenInput)
// CreateCommand 创建 install token,并返回 UI 展示安装命令所需的 URL 与嵌入式脚本。
func (s *InstallTokenService) CreateCommand(ctx context.Context, in InstallCommandInput) (*InstallCommandOutput, error) {
masterURL := strings.TrimRight(strings.TrimSpace(in.MasterURL), "/")
if masterURL == "" {
return nil, apperror.BadRequest("INSTALL_TOKEN_INVALID", "masterURL 必填", nil)
deliveryURL, parseErr := url.Parse(masterURL)
if masterURL == "" || parseErr != nil || (deliveryURL.Scheme != "http" && deliveryURL.Scheme != "https") || deliveryURL.Host == "" || deliveryURL.User != nil || deliveryURL.RawQuery != "" || deliveryURL.Fragment != "" || strings.ContainsAny(masterURL, " \t\r\n\"'`$\\") {
return nil, apperror.BadRequest("INSTALL_TOKEN_INVALID", "masterURL 必须是安全的完整 HTTP(S) 地址", parseErr)
}
agentMasterURL := strings.TrimRight(strings.TrimSpace(in.AgentMasterURL), "/")
if agentMasterURL == "" {
agentMasterURL = masterURL
}
in.AgentMasterURL = agentMasterURL
if err := s.validate(in.InstallTokenInput); err != nil {
return nil, err
}
@@ -144,12 +157,15 @@ func (s *InstallTokenService) CreateCommand(ctx context.Context, in InstallComma
return nil, apperror.New(404, "NODE_NOT_FOUND", "节点不存在", nil)
}
if _, err := renderInstallCommandScript(masterURL, node, &model.AgentInstallToken{
Mode: in.Mode,
Arch: in.Arch,
AgentVer: in.AgentVersion,
DownloadSrc: in.DownloadSrc,
Mode: in.Mode,
Arch: in.Arch,
AgentVer: in.AgentVersion,
DownloadSrc: in.DownloadSrc,
AgentMasterURL: in.AgentMasterURL,
ProxyURL: in.ProxyURL,
CACertFile: in.CACertFile,
}); err != nil {
return nil, err
return nil, apperror.BadRequest("INSTALL_TOKEN_INVALID", "Agent 连接配置无效", err)
}
out, err := s.Create(ctx, in.InstallTokenInput)
if err != nil {
@@ -164,20 +180,24 @@ func (s *InstallTokenService) CreateCommand(ctx context.Context, in InstallComma
ExpiresAt: out.ExpiresAt,
Node: out.Node,
Record: out.Record,
URL: masterURL + "/api/install/" + out.Token,
FallbackURL: masterURL + "/install/" + out.Token,
URL: agentMasterURL + "/api/install/" + out.Token,
FallbackURL: agentMasterURL + "/install/" + out.Token,
ScriptBase64: base64.StdEncoding.EncodeToString([]byte(script)),
}
if out.Record.Mode == model.InstallModeDocker {
result.ComposeURL = masterURL + "/api/install/" + out.Token + "/compose.yml"
result.FallbackComposeURL = masterURL + "/install/" + out.Token + "/compose.yml"
result.ComposeURL = agentMasterURL + "/api/install/" + out.Token + "/compose.yml"
result.FallbackComposeURL = agentMasterURL + "/install/" + out.Token + "/compose.yml"
}
return result, nil
}
func renderInstallCommandScript(masterURL string, node *model.Node, record *model.AgentInstallToken) (string, error) {
agentMasterURL := strings.TrimRight(strings.TrimSpace(record.AgentMasterURL), "/")
if agentMasterURL == "" {
agentMasterURL = masterURL
}
return installscript.RenderScript(installscript.Context{
MasterURL: masterURL,
MasterURL: agentMasterURL,
AgentToken: node.Token,
AgentVersion: record.AgentVer,
Mode: record.Mode,
@@ -185,6 +205,8 @@ func renderInstallCommandScript(masterURL string, node *model.Node, record *mode
DownloadBase: installscript.DownloadBaseFor(record.DownloadSrc),
InstallPrefix: "/opt/backupx-agent",
NodeID: node.ID,
ProxyURL: record.ProxyURL,
CACertFile: record.CACertFile,
})
}
@@ -259,6 +281,9 @@ func (s *InstallTokenService) validate(in InstallTokenInput) error {
return apperror.BadRequest("INSTALL_TOKEN_INVALID",
fmt.Sprintf("ttlSeconds 需在 %d-%d", InstallTokenMinTTL, InstallTokenMaxTTL), nil)
}
if len(in.AgentMasterURL) > 2048 || len(in.ProxyURL) > 2048 || len(in.CACertFile) > 512 {
return apperror.BadRequest("INSTALL_TOKEN_INVALID", "连接配置过长", nil)
}
return nil
}
@@ -2,7 +2,9 @@ package service
import (
"context"
"encoding/base64"
"path/filepath"
"strings"
"testing"
"time"
@@ -177,13 +179,16 @@ func TestInstallTokenServiceCreateCommandBuildsURLsAndScript(t *testing.T) {
out, err := svc.CreateCommand(context.Background(), InstallCommandInput{
InstallTokenInput: InstallTokenInput{
NodeID: node.ID,
Mode: model.InstallModeDocker,
Arch: model.InstallArchAuto,
AgentVersion: "v1.7.0",
DownloadSrc: model.InstallSourceGitHub,
TTLSeconds: 900,
CreatedByID: 1,
NodeID: node.ID,
Mode: model.InstallModeDocker,
Arch: model.InstallArchAuto,
AgentVersion: "v1.7.0",
DownloadSrc: model.InstallSourceGitHub,
TTLSeconds: 900,
CreatedByID: 1,
AgentMasterURL: "http://127.0.0.1:18340",
ProxyURL: "socks5h://127.0.0.1:1080",
CACertFile: "/etc/pki/internal-ca.pem",
},
MasterURL: "https://public.example.com/base",
})
@@ -193,15 +198,66 @@ func TestInstallTokenServiceCreateCommandBuildsURLsAndScript(t *testing.T) {
if out.Token == "" || out.ScriptBase64 == "" {
t.Fatalf("missing token or script: %+v", out)
}
if out.URL != "https://public.example.com/base/api/install/"+out.Token {
if out.URL != "http://127.0.0.1:18340/api/install/"+out.Token {
t.Fatalf("bad url: %s", out.URL)
}
if out.FallbackURL != "https://public.example.com/base/install/"+out.Token {
if out.FallbackURL != "http://127.0.0.1:18340/install/"+out.Token {
t.Fatalf("bad fallback url: %s", out.FallbackURL)
}
if out.ComposeURL != "https://public.example.com/base/api/install/"+out.Token+"/compose.yml" {
if out.ComposeURL != "http://127.0.0.1:18340/api/install/"+out.Token+"/compose.yml" {
t.Fatalf("bad compose url: %s", out.ComposeURL)
}
script, err := base64.StdEncoding.DecodeString(out.ScriptBase64)
if err != nil {
t.Fatal(err)
}
for _, want := range []string{
`MASTER_URL="http://127.0.0.1:18340"`,
`PROXY_URL="socks5h://127.0.0.1:1080"`,
`CA_CERT_FILE="/etc/pki/internal-ca.pem"`,
} {
if !strings.Contains(string(script), want) {
t.Fatalf("script missing %q", want)
}
}
}
func TestInstallTokenServiceRejectsUnsafeConnectionBeforeCreate(t *testing.T) {
db := openInstallTokenTestDB(t)
nodeRepo := repository.NewNodeRepository(db)
node := &model.Node{Name: "restricted", Token: "deadbeefcafebabe0123456789abcdef0123456789abcdef0123456789abcdef"}
if err := nodeRepo.Create(context.Background(), node); err != nil {
t.Fatal(err)
}
tokenRepo := repository.NewAgentInstallTokenRepository(db)
svc := NewInstallTokenService(tokenRepo, nodeRepo)
_, err := svc.CreateCommand(context.Background(), InstallCommandInput{
InstallTokenInput: InstallTokenInput{
NodeID: node.ID, Mode: model.InstallModeSystemd, Arch: model.InstallArchAuto,
AgentVersion: "v2.4.0", DownloadSrc: model.InstallSourceGitHub, TTLSeconds: 900,
ProxyURL: "http://user:pass@proxy.example.com",
},
MasterURL: "https://public.example.com",
})
if err == nil {
t.Fatal("expected proxy credentials to be rejected")
}
count, countErr := tokenRepo.CountCreatedSince(context.Background(), node.ID, time.Now().UTC().Add(-time.Hour))
if countErr != nil || count != 0 {
t.Fatalf("invalid connection created token records: count=%d err=%v", count, countErr)
}
_, err = svc.CreateCommand(context.Background(), InstallCommandInput{
InstallTokenInput: InstallTokenInput{
NodeID: node.ID, Mode: model.InstallModeSystemd, Arch: model.InstallArchAuto,
AgentVersion: "v2.4.0", DownloadSrc: model.InstallSourceGitHub, TTLSeconds: 900,
AgentMasterURL: "https://master.internal",
},
MasterURL: "http://public.example.com/$unsafe",
})
if err == nil {
t.Fatal("expected unsafe public delivery URL to be rejected")
}
}
func TestInstallTokenServiceRateLimit(t *testing.T) {
+84 -8
View File
@@ -4,6 +4,7 @@ import (
"context"
"encoding/json"
"fmt"
"io"
"os"
"path/filepath"
"strings"
@@ -321,6 +322,13 @@ func (s *RestoreService) restoreArtifact(ctx context.Context, record *model.Back
if err != nil {
return fmt.Errorf("创建存储客户端失败:%w", err)
}
if record.BackupKind == model.BackupKindRepository {
logger.Infof("读取 CDC 仓库快照:%s", record.StoragePath)
if err := backup.NewRepositoryStore(s.cipher.Key()).Restore(ctx, provider, record.StoragePath, record.Checksum, spec, logger); err != nil {
return fmt.Errorf("恢复 CDC 仓库快照失败:%w", err)
}
return nil
}
recDir, err := os.MkdirTemp(parentTempDir, fmt.Sprintf("rec-%d-*", record.ID))
if err != nil {
return fmt.Errorf("创建恢复子目录失败:%w", err)
@@ -368,6 +376,9 @@ func (s *RestoreService) buildRestoreChain(ctx context.Context, record *model.Ba
}
func backupKindLabel(kind string) string {
if kind == model.BackupKindRepository {
return "CDC 仓库快照"
}
if kind == model.BackupKindDifferential {
return "差异"
}
@@ -591,15 +602,22 @@ func (s *RestoreService) GetAgentRestoreSpec(ctx context.Context, node *model.No
if target == nil {
return nil, apperror.BadRequest("BACKUP_STORAGE_TARGET_INVALID", "存储目标不存在", nil)
}
configRaw, err := s.cipher.Decrypt(target.ConfigCiphertext)
if err != nil {
return nil, fmt.Errorf("decrypt storage config: %w", err)
}
// 拆开 sourcePaths
sourcePaths := []string{}
if strings.TrimSpace(task.SourcePaths) != "" {
_ = json.Unmarshal([]byte(task.SourcePaths), &sourcePaths)
}
transferMode := storage.TransferModeDirect
if backupRecord.StorageTransferMode == storage.TransferModeMasterRelay {
transferMode = storage.TransferModeMasterRelay
}
var configRaw []byte
if transferMode == storage.TransferModeDirect {
configRaw, err = s.cipher.Decrypt(target.ConfigCiphertext)
if err != nil {
return nil, fmt.Errorf("decrypt storage config: %w", err)
}
}
return &AgentRestoreSpec{
RestoreRecordID: restore.ID,
BackupRecordID: backupRecord.ID,
@@ -618,10 +636,11 @@ func (s *RestoreService) GetAgentRestoreSpec(ctx context.Context, node *model.No
Compression: task.Compression,
Encrypt: task.Encrypt,
Storage: AgentStorageTargetConfig{
ID: target.ID,
Type: target.Type,
Name: target.Name,
Config: json.RawMessage(configRaw),
ID: target.ID,
Type: target.Type,
Name: target.Name,
Config: json.RawMessage(configRaw),
TransferMode: transferMode,
},
StoragePath: backupRecord.StoragePath,
FileName: backupRecord.FileName,
@@ -629,6 +648,63 @@ func (s *RestoreService) GetAgentRestoreSpec(ctx context.Context, node *model.No
}, nil
}
type AgentArtifactDownload struct {
Reader io.ReadCloser
Size int64
}
// DownloadAgentArtifact opens a Master-local object for authenticated streaming
// back to the Agent that owns the restore record.
func (s *RestoreService) DownloadAgentArtifact(ctx context.Context, node *model.Node, restoreID uint) (*AgentArtifactDownload, error) {
if node == nil {
return nil, apperror.Unauthorized("RESTORE_RECORD_FORBIDDEN", "恢复记录不属于当前节点", nil)
}
restore, err := s.restores.FindByID(ctx, restoreID)
if err != nil {
return nil, err
}
if restore == nil {
return nil, apperror.New(404, "RESTORE_RECORD_NOT_FOUND", "恢复记录不存在", nil)
}
if restore.NodeID != node.ID {
return nil, apperror.Unauthorized("RESTORE_RECORD_FORBIDDEN", "恢复记录不属于当前节点", nil)
}
if isRestoreRecordTerminal(restore.Status) {
return nil, apperror.BadRequest("RESTORE_RECORD_TERMINAL", "恢复记录已结束,不能继续下载产物", nil)
}
record, err := s.records.FindByID(ctx, restore.BackupRecordID)
if err != nil {
return nil, err
}
if record == nil {
return nil, apperror.New(404, "BACKUP_RECORD_NOT_FOUND", "源备份记录不存在", nil)
}
target, err := s.targets.FindByID(ctx, record.StorageTargetID)
if err != nil {
return nil, err
}
if target == nil || !strings.EqualFold(target.Type, storage.TypeLocalDisk) || record.StorageTransferMode != storage.TransferModeMasterRelay {
return nil, apperror.BadRequest("AGENT_ARTIFACT_RELAY_UNSUPPORTED", "该存储目标应由 Agent 直接下载", nil)
}
configMap := map[string]any{}
if err := s.cipher.DecryptJSON(target.ConfigCiphertext, &configMap); err != nil {
return nil, fmt.Errorf("decrypt storage config: %w", err)
}
provider, err := s.storageRegistry.Create(ctx, target.Type, configMap)
if err != nil {
return nil, fmt.Errorf("create master relay provider: %w", err)
}
reader, err := provider.Download(ctx, record.StoragePath)
if err != nil {
return nil, fmt.Errorf("open master relay artifact: %w", err)
}
size := record.FileSize
if size <= 0 {
size = -1
}
return &AgentArtifactDownload{Reader: reader, Size: size}, nil
}
// UpdateAgentRestore Agent 回传状态/日志。
func (s *RestoreService) UpdateAgentRestore(ctx context.Context, node *model.Node, restoreID uint, update AgentRestoreUpdate) error {
restore, err := s.restores.FindByID(ctx, restoreID)
@@ -1,8 +1,10 @@
package service
import (
"bytes"
"context"
"encoding/json"
"io"
"os"
"path/filepath"
"strings"
@@ -427,16 +429,27 @@ func TestRestoreServiceAgentRestoreAccessUsesRestoreRecordNode(t *testing.T) {
}
startedAt := time.Now().UTC()
completedAt := startedAt.Add(time.Second)
artifact := []byte("central backup artifact")
storagePath := "file/2026/05/09/remote.tar.gz"
artifactPath := filepath.Join(h.storageDir, filepath.FromSlash(storagePath))
if err := os.MkdirAll(filepath.Dir(artifactPath), 0o755); err != nil {
t.Fatalf("MkdirAll artifact parent: %v", err)
}
if err := os.WriteFile(artifactPath, artifact, 0o600); err != nil {
t.Fatalf("WriteFile artifact: %v", err)
}
backupRecord := &model.BackupRecord{
TaskID: task.ID,
StorageTargetID: task.StorageTargetID,
NodeID: owner.ID,
Status: model.BackupRecordStatusSuccess,
FileName: "remote.tar.gz",
StoragePath: "file/2026/05/09/remote.tar.gz",
Checksum: "0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef",
StartedAt: startedAt,
CompletedAt: &completedAt,
TaskID: task.ID,
StorageTargetID: task.StorageTargetID,
NodeID: owner.ID,
Status: model.BackupRecordStatusSuccess,
FileName: "remote.tar.gz",
StoragePath: storagePath,
FileSize: int64(len(artifact)),
StorageTransferMode: storage.TransferModeMasterRelay,
Checksum: "0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef",
StartedAt: startedAt,
CompletedAt: &completedAt,
}
if err := h.records.Create(ctx, backupRecord); err != nil {
t.Fatalf("Create backup record: %v", err)
@@ -464,6 +477,21 @@ func TestRestoreServiceAgentRestoreAccessUsesRestoreRecordNode(t *testing.T) {
if spec.Checksum != backupRecord.Checksum {
t.Fatalf("expected spec.Checksum=%q, got %q", backupRecord.Checksum, spec.Checksum)
}
if spec.Storage.TransferMode != storage.TransferModeMasterRelay {
t.Fatalf("expected Master relay restore, got %#v", spec.Storage)
}
download, err := h.service.DownloadAgentArtifact(ctx, owner, restore.ID)
if err != nil {
t.Fatalf("DownloadAgentArtifact returned error: %v", err)
}
downloaded, readErr := io.ReadAll(download.Reader)
closeErr := download.Reader.Close()
if readErr != nil || closeErr != nil {
t.Fatalf("read relayed restore artifact: read=%v close=%v", readErr, closeErr)
}
if !bytes.Equal(downloaded, artifact) {
t.Fatalf("relayed restore artifact differs: %q", downloaded)
}
if _, err := h.service.GetAgentRestoreSpec(ctx, other, restore.ID); err == nil {
t.Fatal("expected non-owner node to be forbidden from restore spec")
}
+85 -85
View File
@@ -14,9 +14,9 @@ import (
// TaskExportService 管理备份任务的 JSON 导入 / 导出。
// 用途:
// 1. 集群迁移(旧 Master → 新 Master 的任务配置搬迁)
// 2. 灾备恢复(任务配置本地文件化,Master 宕机后重建)
// 3. 配置审计(版本化 Git 管理 JSON 快照)
// 1. 集群迁移(旧 Master → 新 Master 的任务配置搬迁)
// 2. 灾备恢复(任务配置本地文件化,Master 宕机后重建)
// 3. 配置审计(版本化 Git 管理 JSON 快照)
//
// 出于安全考虑,导出/导入不包含任何敏感字段:
// - 数据库密码(DBPasswordCiphertext):跳过,导入后需人工填补
@@ -40,35 +40,35 @@ func NewTaskExportService(
// ExportedTask 导出格式:按名称引用存储/节点,不含敏感数据。
type ExportedTask struct {
Name string `json:"name"`
Type string `json:"type"`
Enabled bool `json:"enabled"`
CronExpr string `json:"cronExpr,omitempty"`
SourcePath string `json:"sourcePath,omitempty"`
SourcePaths []string `json:"sourcePaths,omitempty"`
ExcludePatterns []string `json:"excludePatterns,omitempty"`
DBHost string `json:"dbHost,omitempty"`
DBPort int `json:"dbPort,omitempty"`
DBUser string `json:"dbUser,omitempty"`
DBName string `json:"dbName,omitempty"`
DBPath string `json:"dbPath,omitempty"`
ExtraConfig map[string]any `json:"extraConfig,omitempty"`
Name string `json:"name"`
Type string `json:"type"`
Enabled bool `json:"enabled"`
CronExpr string `json:"cronExpr,omitempty"`
SourcePath string `json:"sourcePath,omitempty"`
SourcePaths []string `json:"sourcePaths,omitempty"`
ExcludePatterns []string `json:"excludePatterns,omitempty"`
DBHost string `json:"dbHost,omitempty"`
DBPort int `json:"dbPort,omitempty"`
DBUser string `json:"dbUser,omitempty"`
DBName string `json:"dbName,omitempty"`
DBPath string `json:"dbPath,omitempty"`
ExtraConfig map[string]any `json:"extraConfig,omitempty"`
// 按名称引用:导入时按名称查找对应 ID
StorageTargetNames []string `json:"storageTargetNames"`
ReplicationTargetNames []string `json:"replicationTargetNames,omitempty"`
NodeName string `json:"nodeName,omitempty"`
DependsOnTaskNames []string `json:"dependsOnTaskNames,omitempty"`
Tags string `json:"tags,omitempty"`
Compression string `json:"compression,omitempty"`
Encrypt bool `json:"encrypt,omitempty"`
RetentionDays int `json:"retentionDays,omitempty"`
MaxBackups int `json:"maxBackups,omitempty"`
VerifyEnabled bool `json:"verifyEnabled,omitempty"`
VerifyCronExpr string `json:"verifyCronExpr,omitempty"`
VerifyMode string `json:"verifyMode,omitempty"`
SLAHoursRPO int `json:"slaHoursRpo,omitempty"`
AlertOnConsecutiveFails int `json:"alertOnConsecutiveFails,omitempty"`
MaintenanceWindows string `json:"maintenanceWindows,omitempty"`
StorageTargetNames []string `json:"storageTargetNames"`
ReplicationTargetNames []string `json:"replicationTargetNames,omitempty"`
NodeName string `json:"nodeName,omitempty"`
DependsOnTaskNames []string `json:"dependsOnTaskNames,omitempty"`
Tags string `json:"tags,omitempty"`
Compression string `json:"compression,omitempty"`
Encrypt bool `json:"encrypt,omitempty"`
RetentionDays int `json:"retentionDays,omitempty"`
MaxBackups int `json:"maxBackups,omitempty"`
VerifyEnabled bool `json:"verifyEnabled,omitempty"`
VerifyCronExpr string `json:"verifyCronExpr,omitempty"`
VerifyMode string `json:"verifyMode,omitempty"`
SLAHoursRPO int `json:"slaHoursRpo,omitempty"`
AlertOnConsecutiveFails int `json:"alertOnConsecutiveFails,omitempty"`
MaintenanceWindows string `json:"maintenanceWindows,omitempty"`
}
// ExportPayload 导出整体结构,带元信息。
@@ -233,67 +233,67 @@ func (s *TaskExportService) toExported(item *model.BackupTask, targetNames, node
nodeName = nodeNames[item.NodeID]
}
return ExportedTask{
Name: item.Name,
Type: item.Type,
Enabled: item.Enabled,
CronExpr: item.CronExpr,
SourcePath: item.SourcePath,
SourcePaths: sourcePaths,
ExcludePatterns: excludes,
DBHost: item.DBHost,
DBPort: item.DBPort,
DBUser: item.DBUser,
DBName: item.DBName,
DBPath: item.DBPath,
ExtraConfig: extra,
StorageTargetNames: storageNames,
ReplicationTargetNames: replicationNames,
NodeName: nodeName,
DependsOnTaskNames: dependsOnNames,
Tags: item.Tags,
Compression: item.Compression,
Encrypt: item.Encrypt,
RetentionDays: item.RetentionDays,
MaxBackups: item.MaxBackups,
VerifyEnabled: item.VerifyEnabled,
VerifyCronExpr: item.VerifyCronExpr,
VerifyMode: item.VerifyMode,
SLAHoursRPO: item.SLAHoursRPO,
Name: item.Name,
Type: item.Type,
Enabled: item.Enabled,
CronExpr: item.CronExpr,
SourcePath: item.SourcePath,
SourcePaths: sourcePaths,
ExcludePatterns: excludes,
DBHost: item.DBHost,
DBPort: item.DBPort,
DBUser: item.DBUser,
DBName: item.DBName,
DBPath: item.DBPath,
ExtraConfig: extra,
StorageTargetNames: storageNames,
ReplicationTargetNames: replicationNames,
NodeName: nodeName,
DependsOnTaskNames: dependsOnNames,
Tags: item.Tags,
Compression: item.Compression,
Encrypt: item.Encrypt,
RetentionDays: item.RetentionDays,
MaxBackups: item.MaxBackups,
VerifyEnabled: item.VerifyEnabled,
VerifyCronExpr: item.VerifyCronExpr,
VerifyMode: item.VerifyMode,
SLAHoursRPO: item.SLAHoursRPO,
AlertOnConsecutiveFails: item.AlertOnConsecutiveFails,
MaintenanceWindows: item.MaintenanceWindows,
MaintenanceWindows: item.MaintenanceWindows,
}
}
func (s *TaskExportService) toUpsertInput(t ExportedTask, targetsByName, nodesByName map[string]uint, deps []uint) BackupTaskUpsertInput {
return BackupTaskUpsertInput{
Name: t.Name,
Type: t.Type,
Enabled: t.Enabled,
CronExpr: t.CronExpr,
SourcePath: t.SourcePath,
SourcePaths: t.SourcePaths,
ExcludePatterns: t.ExcludePatterns,
DBHost: t.DBHost,
DBPort: t.DBPort,
DBUser: t.DBUser,
DBName: t.DBName,
DBPath: t.DBPath,
ExtraConfig: t.ExtraConfig,
StorageTargetIDs: idsFromNames(t.StorageTargetNames, targetsByName),
ReplicationTargetIDs: idsFromNames(t.ReplicationTargetNames, targetsByName),
NodeID: nodesByName[t.NodeName],
Tags: t.Tags,
Compression: t.Compression,
Encrypt: t.Encrypt,
RetentionDays: t.RetentionDays,
MaxBackups: t.MaxBackups,
VerifyEnabled: t.VerifyEnabled,
VerifyCronExpr: t.VerifyCronExpr,
VerifyMode: t.VerifyMode,
SLAHoursRPO: t.SLAHoursRPO,
Name: t.Name,
Type: t.Type,
Enabled: t.Enabled,
CronExpr: t.CronExpr,
SourcePath: t.SourcePath,
SourcePaths: t.SourcePaths,
ExcludePatterns: t.ExcludePatterns,
DBHost: t.DBHost,
DBPort: t.DBPort,
DBUser: t.DBUser,
DBName: t.DBName,
DBPath: t.DBPath,
ExtraConfig: t.ExtraConfig,
StorageTargetIDs: idsFromNames(t.StorageTargetNames, targetsByName),
ReplicationTargetIDs: idsFromNames(t.ReplicationTargetNames, targetsByName),
NodeID: nodesByName[t.NodeName],
Tags: t.Tags,
Compression: t.Compression,
Encrypt: t.Encrypt,
RetentionDays: t.RetentionDays,
MaxBackups: t.MaxBackups,
VerifyEnabled: t.VerifyEnabled,
VerifyCronExpr: t.VerifyCronExpr,
VerifyMode: t.VerifyMode,
SLAHoursRPO: t.SLAHoursRPO,
AlertOnConsecutiveFails: t.AlertOnConsecutiveFails,
MaintenanceWindows: t.MaintenanceWindows,
DependsOnTaskIDs: deps,
MaintenanceWindows: t.MaintenanceWindows,
DependsOnTaskIDs: deps,
}
}
@@ -299,6 +299,20 @@ func (s *VerificationService) executeLocally(ctx context.Context, verID uint, ta
logger.Errorf("创建存储客户端失败:%v", err)
return
}
if backupRecord.BackupKind == model.BackupKindRepository {
logger.Infof("验证 CDC 仓库快照及全部引用块:%s", backupRecord.StoragePath)
report, verifyErr := backup.NewRepositoryStore(s.cipher.Key()).Verify(ctx, provider, backupRecord.StoragePath, backupRecord.Checksum)
if verifyErr != nil {
errMessage = verifyErr.Error()
summary = "CDC 仓库完整性校验失败"
logger.Errorf("验证未通过:%v", verifyErr)
return
}
status = model.VerificationRecordStatusSuccess
summary = fmt.Sprintf("CDC 仓库完整性校验通过:%d 个条目、%d 个唯一块、%d bytes", report.Entries, report.Chunks, report.Bytes)
logger.Infof("%s", summary)
return
}
fileName := backupRecord.FileName
if strings.TrimSpace(fileName) == "" {
fileName = filepath.Base(backupRecord.StoragePath)