diff --git a/backend/internal/repository/scheduler_cache.go b/backend/internal/repository/scheduler_cache.go index ab01a863..ec8c72dc 100644 --- a/backend/internal/repository/scheduler_cache.go +++ b/backend/internal/repository/scheduler_cache.go @@ -548,6 +548,12 @@ func filterSchedulerExtra(extra map[string]any) map[string]any { "openai_ws_force_http", "openai_responses_mode", "openai_responses_supported", + "codex_5h_used_percent", + "codex_7d_used_percent", + "auto_pause_5h_threshold", + "auto_pause_7d_threshold", + "auto_pause_5h_limit", + "auto_pause_7d_limit", } filtered := make(map[string]any) for _, key := range keys { diff --git a/backend/internal/repository/scheduler_cache_unit_test.go b/backend/internal/repository/scheduler_cache_unit_test.go index 86de87c7..fabc6bad 100644 --- a/backend/internal/repository/scheduler_cache_unit_test.go +++ b/backend/internal/repository/scheduler_cache_unit_test.go @@ -75,3 +75,22 @@ func TestBuildSchedulerMetadataAccount_KeepsSlimGroupMembership(t *testing.T) { require.Equal(t, int64(11), got.AccountGroups[1].GroupID) require.Nil(t, got.Groups) } + +func TestBuildSchedulerMetadataAccount_KeepsQuotaAutoPauseFields(t *testing.T) { + account := service.Account{ + ID: 88, + Extra: map[string]any{ + "codex_5h_used_percent": 12.34, + "codex_7d_used_percent": 56.78, + "auto_pause_5h_threshold": 0.95, + "auto_pause_7d_threshold": 0.96, + }, + } + + got := buildSchedulerMetadataAccount(account) + + require.Equal(t, 12.34, got.Extra["codex_5h_used_percent"]) + require.Equal(t, 56.78, got.Extra["codex_7d_used_percent"]) + require.Equal(t, 0.95, got.Extra["auto_pause_5h_threshold"]) + require.Equal(t, 0.96, got.Extra["auto_pause_7d_threshold"]) +} diff --git a/backend/internal/service/openai_account_scheduler.go b/backend/internal/service/openai_account_scheduler.go index 1eca08b1..fd28fa86 100644 --- a/backend/internal/service/openai_account_scheduler.go +++ b/backend/internal/service/openai_account_scheduler.go @@ -370,7 +370,6 @@ func (s *defaultOpenAIAccountScheduler) selectBySessionHash( _ = s.service.deleteStickySessionAccountID(ctx, req.GroupID, sessionHash) return nil, nil } - result, acquireErr := s.service.tryAcquireAccountSlot(ctx, accountID, account.Concurrency) if acquireErr == nil && result != nil && result.Acquired { _ = s.service.refreshStickySessionTTL(ctx, req.GroupID, sessionHash, s.service.openAIWSSessionStickyTTL()) @@ -1154,6 +1153,7 @@ func (s *OpenAIGatewayService) selectAccountWithScheduler( requiredImageCapability OpenAIImagesCapability, requireCompact bool, ) (*AccountSelectionResult, OpenAIAccountScheduleDecision, error) { + ctx = s.withOpenAIQuotaAutoPauseContext(ctx) decision := OpenAIAccountScheduleDecision{} scheduler := s.getOpenAIAccountScheduler(ctx) if scheduler == nil { diff --git a/backend/internal/service/openai_account_scheduler_test.go b/backend/internal/service/openai_account_scheduler_test.go index fedf7e9c..37810870 100644 --- a/backend/internal/service/openai_account_scheduler_test.go +++ b/backend/internal/service/openai_account_scheduler_test.go @@ -691,6 +691,117 @@ func TestOpenAIGatewayService_SelectAccountWithScheduler_SessionStickyRateLimite require.Equal(t, openAIAccountScheduleLayerLoadBalance, decision.Layer) } +func TestOpenAIGatewayService_SelectAccountForModelWithExclusions_AutoPauseBy5hThreshold(t *testing.T) { + ctx := context.Background() + primary := Account{ + ID: 35001, + Platform: PlatformOpenAI, + Type: AccountTypeAPIKey, + Status: StatusActive, + Schedulable: true, + Concurrency: 1, + Priority: 0, + Extra: map[string]any{ + "codex_5h_used_percent": 95.0, + "auto_pause_5h_threshold": 0.95, + "auto_pause_5h_limit": 100, + }, + } + secondary := Account{ID: 35002, Platform: PlatformOpenAI, Type: AccountTypeAPIKey, Status: StatusActive, Schedulable: true, Concurrency: 1, Priority: 5} + svc := &OpenAIGatewayService{accountRepo: schedulerTestOpenAIAccountRepo{accounts: []Account{primary, secondary}}, cfg: &config.Config{}} + + account, err := svc.SelectAccountForModelWithExclusions(ctx, nil, "", "gpt-5.1", nil) + require.NoError(t, err) + require.NotNil(t, account) + require.Equal(t, int64(35002), account.ID) +} + +func TestOpenAIGatewayService_SelectAccountForModelWithExclusions_AllowsBelow5hThreshold(t *testing.T) { + ctx := context.Background() + primary := Account{ + ID: 35101, + Platform: PlatformOpenAI, + Type: AccountTypeAPIKey, + Status: StatusActive, + Schedulable: true, + Concurrency: 1, + Priority: 0, + Extra: map[string]any{ + "codex_5h_used_percent": 80.0, + "auto_pause_5h_threshold": 0.95, + "auto_pause_5h_limit": 100, + }, + } + secondary := Account{ID: 35102, Platform: PlatformOpenAI, Type: AccountTypeAPIKey, Status: StatusActive, Schedulable: true, Concurrency: 1, Priority: 5} + svc := &OpenAIGatewayService{accountRepo: schedulerTestOpenAIAccountRepo{accounts: []Account{primary, secondary}}, cfg: &config.Config{}} + + account, err := svc.SelectAccountForModelWithExclusions(ctx, nil, "", "gpt-5.1", nil) + require.NoError(t, err) + require.NotNil(t, account) + require.Equal(t, int64(35101), account.ID) +} + +func TestOpenAIGatewayService_SelectAccountForModelWithExclusions_AutoPauseBy7dThreshold(t *testing.T) { + ctx := context.Background() + primary := Account{ + ID: 35201, + Platform: PlatformOpenAI, + Type: AccountTypeAPIKey, + Status: StatusActive, + Schedulable: true, + Concurrency: 1, + Priority: 0, + Extra: map[string]any{ + "codex_7d_used_percent": 95.0, + "auto_pause_7d_threshold": 0.95, + "auto_pause_7d_limit": 200, + }, + } + secondary := Account{ID: 35202, Platform: PlatformOpenAI, Type: AccountTypeAPIKey, Status: StatusActive, Schedulable: true, Concurrency: 1, Priority: 5} + svc := &OpenAIGatewayService{accountRepo: schedulerTestOpenAIAccountRepo{accounts: []Account{primary, secondary}}, cfg: &config.Config{}} + + account, err := svc.SelectAccountForModelWithExclusions(ctx, nil, "", "gpt-5.1", nil) + require.NoError(t, err) + require.NotNil(t, account) + require.Equal(t, int64(35202), account.ID) +} + +func TestOpenAIGatewayService_SelectAccountForModelWithExclusions_UnconfiguredThresholdKeepsLegacyBehavior(t *testing.T) { + ctx := context.Background() + primary := Account{ID: 35301, Platform: PlatformOpenAI, Type: AccountTypeAPIKey, Status: StatusActive, Schedulable: true, Concurrency: 1, Priority: 0, Extra: map[string]any{"codex_5h_used_percent": 99.0, "codex_7d_used_percent": 99.0}} + secondary := Account{ID: 35302, Platform: PlatformOpenAI, Type: AccountTypeAPIKey, Status: StatusActive, Schedulable: true, Concurrency: 1, Priority: 5} + svc := &OpenAIGatewayService{accountRepo: schedulerTestOpenAIAccountRepo{accounts: []Account{primary, secondary}}, cfg: &config.Config{}} + + account, err := svc.SelectAccountForModelWithExclusions(ctx, nil, "", "gpt-5.1", nil) + require.NoError(t, err) + require.NotNil(t, account) + require.Equal(t, int64(35301), account.ID) +} + +func TestOpenAIGatewayService_SelectAccountForModelWithExclusions_UsesGlobalDefaultThreshold(t *testing.T) { + ctx := withOpenAIQuotaAutoPauseSettings(context.Background(), OpsOpenAIAccountQuotaAutoPauseSettings{DefaultThreshold5h: 0.95}) + primary := Account{ + ID: 35401, + Platform: PlatformOpenAI, + Type: AccountTypeAPIKey, + Status: StatusActive, + Schedulable: true, + Concurrency: 1, + Priority: 0, + Extra: map[string]any{ + "codex_5h_used_percent": 95.0, + "auto_pause_5h_limit": 100, + }, + } + secondary := Account{ID: 35402, Platform: PlatformOpenAI, Type: AccountTypeAPIKey, Status: StatusActive, Schedulable: true, Concurrency: 1, Priority: 5} + svc := &OpenAIGatewayService{accountRepo: schedulerTestOpenAIAccountRepo{accounts: []Account{primary, secondary}}, cfg: &config.Config{}} + + account, err := svc.SelectAccountForModelWithExclusions(ctx, nil, "", "gpt-5.1", nil) + require.NoError(t, err) + require.NotNil(t, account) + require.Equal(t, int64(35402), account.ID) +} + func TestOpenAIGatewayService_SelectAccountForModelWithExclusions_SkipsFreshlyRateLimitedSnapshotCandidate(t *testing.T) { ctx := context.Background() groupID := int64(10102) diff --git a/backend/internal/service/openai_gateway_service.go b/backend/internal/service/openai_gateway_service.go index 623a64e9..268f985c 100644 --- a/backend/internal/service/openai_gateway_service.go +++ b/backend/internal/service/openai_gateway_service.go @@ -1290,7 +1290,7 @@ func (s *OpenAIGatewayService) SelectAccountForModel(ctx context.Context, groupI // SelectAccountForModelWithExclusions selects an account supporting the requested model while excluding specified accounts. // SelectAccountForModelWithExclusions 选择支持指定模型的账号，同时排除指定的账号。 func (s *OpenAIGatewayService) SelectAccountForModelWithExclusions(ctx context.Context, groupID *int64, sessionHash string, requestedModel string, excludedIDs map[int64]struct{}) (*Account, error) { - return s.selectAccountForModelWithExclusions(ctx, groupID, sessionHash, requestedModel, excludedIDs, false, 0, "") + return s.selectAccountForModelWithExclusions(s.withOpenAIQuotaAutoPauseContext(ctx), groupID, sessionHash, requestedModel, excludedIDs, false, 0, "") } // noAvailableOpenAISelectionError builds the standard "no account available" error @@ -1327,6 +1327,18 @@ func isOpenAIAccountEligibleForRequest(ctx context.Context, account *Account, re if account == nil || !account.IsOpenAI() || !account.IsSchedulableForModelWithContext(ctx, requestedModel) { return false } + if paused, reason := shouldAutoPauseOpenAIAccountByQuota(ctx, account); paused { + slog.Info("account_auto_paused_by_quota", + "account_id", account.ID, + "usage_5h_percent", readOpenAIQuotaUsedPercent(account.Extra, "5h"), + "usage_7d_percent", readOpenAIQuotaUsedPercent(account.Extra, "7d"), + "threshold_type", reason.window, + "threshold", reason.threshold, + "limit", reason.limit, + "utilization", reason.utilization, + ) + return false + } if requestedModel != "" && !account.IsModelSupported(requestedModel) { return false } @@ -1339,6 +1351,134 @@ func isOpenAIAccountEligibleForRequest(ctx context.Context, account *Account, re return true } +type openAIQuotaAutoPauseDecision struct { + window string + threshold float64 + limit float64 + utilization float64 +} + +func shouldAutoPauseOpenAIAccountByQuota(ctx context.Context, account *Account) (bool, openAIQuotaAutoPauseDecision) { + if account == nil || !account.IsOpenAI() { + return false, openAIQuotaAutoPauseDecision{} + } + threshold5h, threshold7d := resolveOpenAIQuotaAutoPauseThresholds(ctx, account) + if threshold5h > 0 { + if utilization, limit, ok := resolveOpenAIQuotaUtilization(account.Extra, "5h"); ok { + if utilization >= threshold5h { + return true, openAIQuotaAutoPauseDecision{window: "5h", threshold: threshold5h, limit: limit, utilization: utilization} + } + } + } + if threshold7d > 0 { + if utilization, limit, ok := resolveOpenAIQuotaUtilization(account.Extra, "7d"); ok { + if utilization >= threshold7d { + return true, openAIQuotaAutoPauseDecision{window: "7d", threshold: threshold7d, limit: limit, utilization: utilization} + } + } + } + return false, openAIQuotaAutoPauseDecision{} +} + +func resolveOpenAIQuotaAutoPauseThresholds(ctx context.Context, account *Account) (float64, float64) { + threshold5h, _ := resolveAccountExtraNumber(account.Extra, "auto_pause_5h_threshold") + threshold7d, _ := resolveAccountExtraNumber(account.Extra, "auto_pause_7d_threshold") + threshold5h = clamp01(threshold5h) + threshold7d = clamp01(threshold7d) + if threshold5h > 0 && threshold7d > 0 { + return threshold5h, threshold7d + } + settings := openAIQuotaAutoPauseSettingsFromContext(ctx) + if threshold5h <= 0 { + threshold5h = clamp01(settings.DefaultThreshold5h) + } + if threshold7d <= 0 { + threshold7d = clamp01(settings.DefaultThreshold7d) + } + return threshold5h, threshold7d +} + +func resolveAccountExtraNumber(extra map[string]any, keys ...string) (float64, bool) { + if len(extra) == 0 { + return 0, false + } + for _, key := range keys { + value, ok := extra[key] + if !ok || value == nil { + continue + } + switch v := value.(type) { + case float64: + return v, true + case float32: + return float64(v), true + case int: + return float64(v), true + case int64: + return float64(v), true + case json.Number: + parsed, err := v.Float64() + if err == nil { + return parsed, true + } + case string: + parsed, err := strconv.ParseFloat(strings.TrimSpace(v), 64) + if err == nil { + return parsed, true + } + } + } + return 0, false +} + +func resolveOpenAIQuotaUtilization(extra map[string]any, window string) (float64, float64, bool) { + limitKeys := []string{"auto_pause_" + window + "_limit", "quota_" + window + "_limit", window + "_limit"} + if limit, ok := resolveAccountExtraNumber(extra, limitKeys...); ok && limit > 0 { + if usage, ok := resolveAccountExtraNumber(extra, "usage_"+window); ok && usage >= 0 { + return usage / limit, limit, true + } + } + usedPercent := readOpenAIQuotaUsedPercent(extra, window) + if usedPercent <= 0 { + return 0, 0, false + } + return usedPercent / 100, 100, true +} + +func readOpenAIQuotaUsedPercent(extra map[string]any, window string) float64 { + if len(extra) == 0 { + return 0 + } + if value, ok := resolveAccountExtraNumber(extra, "codex_"+window+"_used_percent"); ok { + return value + } + return 0 +} + +type openAIQuotaAutoPauseCtxKey struct{} + +func withOpenAIQuotaAutoPauseSettings(ctx context.Context, settings OpsOpenAIAccountQuotaAutoPauseSettings) context.Context { + if ctx == nil { + ctx = context.Background() + } + return context.WithValue(ctx, openAIQuotaAutoPauseCtxKey{}, settings) +} + +func openAIQuotaAutoPauseSettingsFromContext(ctx context.Context) OpsOpenAIAccountQuotaAutoPauseSettings { + if ctx == nil { + return OpsOpenAIAccountQuotaAutoPauseSettings{} + } + settings, _ := ctx.Value(openAIQuotaAutoPauseCtxKey{}).(OpsOpenAIAccountQuotaAutoPauseSettings) + return settings +} + +func (s *OpenAIGatewayService) withOpenAIQuotaAutoPauseContext(ctx context.Context) context.Context { + if s == nil || s.settingService == nil { + return ctx + } + return withOpenAIQuotaAutoPauseSettings(ctx, s.settingService.GetOpenAIQuotaAutoPauseSettings(ctx)) +} + // prioritizeOpenAICompactAccounts re-orders a slice so that accounts with known // compact support are tried first, followed by unknown, then explicitly unsupported. // The relative order within each tier is preserved. @@ -1587,7 +1727,7 @@ func (s *OpenAIGatewayService) isBetterAccount(candidate, current *Account) bool // SelectAccountWithLoadAwareness selects an account with load-awareness and wait plan. func (s *OpenAIGatewayService) SelectAccountWithLoadAwareness(ctx context.Context, groupID *int64, sessionHash string, requestedModel string, excludedIDs map[int64]struct{}) (*AccountSelectionResult, error) { - return s.selectAccountWithLoadAwareness(ctx, groupID, sessionHash, requestedModel, excludedIDs, false, "") + return s.selectAccountWithLoadAwareness(s.withOpenAIQuotaAutoPauseContext(ctx), groupID, sessionHash, requestedModel, excludedIDs, false, "") } func (s *OpenAIGatewayService) selectAccountWithLoadAwareness(ctx context.Context, groupID *int64, sessionHash string, requestedModel string, excludedIDs map[int64]struct{}, requireCompact bool, requiredCapability OpenAIEndpointCapability) (*AccountSelectionResult, error) { diff --git a/backend/internal/service/ops_settings.go b/backend/internal/service/ops_settings.go index 68c1d9dd..23e92e5a 100644 --- a/backend/internal/service/ops_settings.go +++ b/backend/internal/service/ops_settings.go @@ -369,6 +369,7 @@ func defaultOpsAdvancedSettings() *OpsAdvancedSettings { Aggregation: OpsAggregationSettings{ AggregationEnabled: false, }, + OpenAIAccountQuotaAutoPause: OpsOpenAIAccountQuotaAutoPauseSettings{}, IgnoreCountTokensErrors: true, // count_tokens 404 是预期行为，默认忽略 IgnoreContextCanceled: true, // Default to true - client disconnects are not errors IgnoreNoAvailableAccounts: false, // Default to false - this is a real routing issue @@ -384,6 +385,8 @@ func normalizeOpsAdvancedSettings(cfg *OpsAdvancedSettings) { if cfg == nil { return } + cfg.OpenAIAccountQuotaAutoPause.DefaultThreshold5h = clampOpsQuotaAutoPauseThreshold(cfg.OpenAIAccountQuotaAutoPause.DefaultThreshold5h) + cfg.OpenAIAccountQuotaAutoPause.DefaultThreshold7d = clampOpsQuotaAutoPauseThreshold(cfg.OpenAIAccountQuotaAutoPause.DefaultThreshold7d) cfg.DataRetention.CleanupSchedule = strings.TrimSpace(cfg.DataRetention.CleanupSchedule) if cfg.DataRetention.CleanupSchedule == "" { cfg.DataRetention.CleanupSchedule = opsCleanupDefaultSchedule @@ -405,6 +408,16 @@ func normalizeOpsAdvancedSettings(cfg *OpsAdvancedSettings) { } } +func clampOpsQuotaAutoPauseThreshold(value float64) float64 { + if value <= 0 { + return 0 + } + if value > 1 { + return 1 + } + return value +} + func validateOpsAdvancedSettings(cfg *OpsAdvancedSettings) error { if cfg == nil { return errors.New("invalid config") @@ -477,6 +490,12 @@ func (s *OpsService) UpdateOpsAdvancedSettings(ctx context.Context, cfg *OpsAdva if err := s.settingRepo.Set(ctx, SettingKeyOpsAdvancedSettings, string(raw)); err != nil { return nil, err } + cacheKey := openAIQuotaAutoPauseSettingsCacheKey(s.settingRepo) + openAIQuotaAutoPauseSettingsSF.Forget(cacheKey) + storeOpenAIQuotaAutoPauseSettingsCache(s.settingRepo, &cachedOpenAIQuotaAutoPauseSettings{ + settings: cfg.OpenAIAccountQuotaAutoPause, + expiresAt: time.Now().Add(openAIQuotaAutoPauseSettingsCacheTTL).UnixNano(), + }) // notify cleanup service to reload schedule/enabled. if s.cleanupReloader != nil { diff --git a/backend/internal/service/ops_settings_advanced_test.go b/backend/internal/service/ops_settings_advanced_test.go index 06cc545b..d8598fe0 100644 --- a/backend/internal/service/ops_settings_advanced_test.go +++ b/backend/internal/service/ops_settings_advanced_test.go @@ -4,6 +4,8 @@ import ( "context" "encoding/json" "testing" + + "github.com/Wei-Shaw/sub2api/internal/config" ) func TestGetOpsAdvancedSettings_DefaultHidesOpenAITokenStats(t *testing.T) { @@ -95,3 +97,17 @@ func TestGetOpsAdvancedSettings_BackfillsNewDisplayFlagsFromDefaults(t *testing. t.Fatalf("DisplayAlertEvents = false, want true default backfill") } } + +func TestGetOpenAIQuotaAutoPauseSettings_ReadsDefaultsFromOpsAdvancedSettings(t *testing.T) { + repo := newRuntimeSettingRepoStub() + repo.values[SettingKeyOpsAdvancedSettings] = `{"openai_account_quota_auto_pause":{"default_threshold_5h":0.95,"default_threshold_7d":0.9}}` + svc := NewSettingService(repo, &config.Config{}) + + settings := svc.GetOpenAIQuotaAutoPauseSettings(context.Background()) + if settings.DefaultThreshold5h != 0.95 { + t.Fatalf("DefaultThreshold5h = %v, want 0.95", settings.DefaultThreshold5h) + } + if settings.DefaultThreshold7d != 0.9 { + t.Fatalf("DefaultThreshold7d = %v, want 0.9", settings.DefaultThreshold7d) + } +} diff --git a/backend/internal/service/ops_settings_models.go b/backend/internal/service/ops_settings_models.go index fa18b05f..4d459e2b 100644 --- a/backend/internal/service/ops_settings_models.go +++ b/backend/internal/service/ops_settings_models.go @@ -92,17 +92,23 @@ type OpsAlertRuntimeSettings struct { // OpsAdvancedSettings stores advanced ops configuration (data retention, aggregation). type OpsAdvancedSettings struct { - DataRetention OpsDataRetentionSettings `json:"data_retention"` - Aggregation OpsAggregationSettings `json:"aggregation"` - IgnoreCountTokensErrors bool `json:"ignore_count_tokens_errors"` - IgnoreContextCanceled bool `json:"ignore_context_canceled"` - IgnoreNoAvailableAccounts bool `json:"ignore_no_available_accounts"` - IgnoreInvalidApiKeyErrors bool `json:"ignore_invalid_api_key_errors"` - IgnoreInsufficientBalanceErrors bool `json:"ignore_insufficient_balance_errors"` - DisplayOpenAITokenStats bool `json:"display_openai_token_stats"` - DisplayAlertEvents bool `json:"display_alert_events"` - AutoRefreshEnabled bool `json:"auto_refresh_enabled"` - AutoRefreshIntervalSec int `json:"auto_refresh_interval_seconds"` + DataRetention OpsDataRetentionSettings `json:"data_retention"` + Aggregation OpsAggregationSettings `json:"aggregation"` + OpenAIAccountQuotaAutoPause OpsOpenAIAccountQuotaAutoPauseSettings `json:"openai_account_quota_auto_pause"` + IgnoreCountTokensErrors bool `json:"ignore_count_tokens_errors"` + IgnoreContextCanceled bool `json:"ignore_context_canceled"` + IgnoreNoAvailableAccounts bool `json:"ignore_no_available_accounts"` + IgnoreInvalidApiKeyErrors bool `json:"ignore_invalid_api_key_errors"` + IgnoreInsufficientBalanceErrors bool `json:"ignore_insufficient_balance_errors"` + DisplayOpenAITokenStats bool `json:"display_openai_token_stats"` + DisplayAlertEvents bool `json:"display_alert_events"` + AutoRefreshEnabled bool `json:"auto_refresh_enabled"` + AutoRefreshIntervalSec int `json:"auto_refresh_interval_seconds"` +} + +type OpsOpenAIAccountQuotaAutoPauseSettings struct { + DefaultThreshold5h float64 `json:"default_threshold_5h"` + DefaultThreshold7d float64 `json:"default_threshold_7d"` } type OpsDataRetentionSettings struct { diff --git a/backend/internal/service/setting_service.go b/backend/internal/service/setting_service.go index 08c0d045..40e98b88 100644 --- a/backend/internal/service/setting_service.go +++ b/backend/internal/service/setting_service.go @@ -14,6 +14,7 @@ import ( "sort" "strconv" "strings" + "sync" "sync/atomic" "time" @@ -137,6 +138,11 @@ type cachedOpenAICodexUserAgent struct { expiresAt int64 // unix nano } +type cachedOpenAIQuotaAutoPauseSettings struct { + settings OpsOpenAIAccountQuotaAutoPauseSettings + expiresAt int64 +} + const openAICodexUserAgentCacheTTL = 60 * time.Second const openAICodexUserAgentErrorTTL = 5 * time.Second const openAICodexUserAgentDBTimeout = 5 * time.Second @@ -152,6 +158,33 @@ const openAIAllowCodexPluginCacheTTL = 60 * time.Second const openAIAllowCodexPluginErrorTTL = 5 * time.Second const openAIAllowCodexPluginDBTimeout = 5 * time.Second +const openAIQuotaAutoPauseSettingsCacheTTL = 60 * time.Second +const openAIQuotaAutoPauseSettingsErrorTTL = 5 * time.Second +const openAIQuotaAutoPauseSettingsDBTimeout = 5 * time.Second + +var openAIQuotaAutoPauseSettingsCache sync.Map // map[string]*cachedOpenAIQuotaAutoPauseSettings +var openAIQuotaAutoPauseSettingsSF singleflight.Group + +func openAIQuotaAutoPauseSettingsCacheKey(repo SettingRepository) string { + if repo == nil { + return "nil" + } + return fmt.Sprintf("%T:%p", repo, repo) +} + +func loadOpenAIQuotaAutoPauseSettingsCache(repo SettingRepository) (*cachedOpenAIQuotaAutoPauseSettings, bool) { + value, ok := openAIQuotaAutoPauseSettingsCache.Load(openAIQuotaAutoPauseSettingsCacheKey(repo)) + if !ok || value == nil { + return nil, false + } + cached, ok := value.(*cachedOpenAIQuotaAutoPauseSettings) + return cached, ok && cached != nil +} + +func storeOpenAIQuotaAutoPauseSettingsCache(repo SettingRepository, cached *cachedOpenAIQuotaAutoPauseSettings) { + openAIQuotaAutoPauseSettingsCache.Store(openAIQuotaAutoPauseSettingsCacheKey(repo), cached) +} + // DefaultSubscriptionGroupReader validates group references used by default subscriptions. type DefaultSubscriptionGroupReader interface { GetByID(ctx context.Context, id int64) (*Group, error) @@ -2027,6 +2060,9 @@ func (s *SettingService) refreshCachedSettings(settings *SystemSettings) { enabled: settings.OpenAIAdvancedSchedulerEnabled, expiresAt: time.Now().Add(openAIAdvancedSchedulerSettingCacheTTL).UnixNano(), }) + cacheKey := openAIQuotaAutoPauseSettingsCacheKey(s.settingRepo) + openAIQuotaAutoPauseSettingsSF.Forget(cacheKey) + openAIQuotaAutoPauseSettingsCache.Delete(cacheKey) if s.cfg != nil { s.cfg.SetTrustForwardedIPForAPIKeyACL(settings.APIKeyACLTrustForwardedIP) } @@ -4448,6 +4484,51 @@ func (s *SettingService) GetClaudeCodeVersionBounds(ctx context.Context) (min, m return b.min, b.max } +func (s *SettingService) GetOpenAIQuotaAutoPauseSettings(ctx context.Context) OpsOpenAIAccountQuotaAutoPauseSettings { + if cached, ok := loadOpenAIQuotaAutoPauseSettingsCache(s.settingRepo); ok { + if time.Now().UnixNano() < cached.expiresAt { + return cached.settings + } + } + + cacheKey := openAIQuotaAutoPauseSettingsCacheKey(s.settingRepo) + result, _, _ := openAIQuotaAutoPauseSettingsSF.Do(cacheKey, func() (any, error) { + if cached, ok := loadOpenAIQuotaAutoPauseSettingsCache(s.settingRepo); ok { + if time.Now().UnixNano() < cached.expiresAt { + return cached.settings, nil + } + } + + settings := OpsOpenAIAccountQuotaAutoPauseSettings{} + ttl := openAIQuotaAutoPauseSettingsCacheTTL + if s != nil && s.settingRepo != nil { + dbCtx, cancel := context.WithTimeout(context.WithoutCancel(ctx), openAIQuotaAutoPauseSettingsDBTimeout) + defer cancel() + raw, err := s.settingRepo.GetValue(dbCtx, SettingKeyOpsAdvancedSettings) + if err == nil { + cfg := defaultOpsAdvancedSettings() + if strings.TrimSpace(raw) != "" { + if jsonErr := json.Unmarshal([]byte(raw), cfg); jsonErr == nil { + normalizeOpsAdvancedSettings(cfg) + } + } + settings = cfg.OpenAIAccountQuotaAutoPause + } else { + ttl = openAIQuotaAutoPauseSettingsErrorTTL + } + } + + storeOpenAIQuotaAutoPauseSettingsCache(s.settingRepo, &cachedOpenAIQuotaAutoPauseSettings{ + settings: settings, + expiresAt: time.Now().Add(ttl).UnixNano(), + }) + return settings, nil + }) + + settings, _ := result.(OpsOpenAIAccountQuotaAutoPauseSettings) + return settings +} + // GetRectifierSettings 获取请求整流器配置 func (s *SettingService) GetRectifierSettings(ctx context.Context) (*RectifierSettings, error) { value, err := s.settingRepo.GetValue(ctx, SettingKeyRectifierSettings) diff --git a/frontend/src/components/account/EditAccountModal.vue b/frontend/src/components/account/EditAccountModal.vue index 515ba58f..470c0bfb 100644 --- a/frontend/src/components/account/EditAccountModal.vue +++ b/frontend/src/components/account/EditAccountModal.vue @@ -1787,6 +1787,38 @@ +
+
+ + +

{{ t('admin.accounts.autoPauseThresholdHint') }}

+
+
+ + +

{{ t('admin.accounts.autoPauseThresholdHint') }}

+
+
+
([]) const customErrorCodeInput = ref(null) const interceptWarmupRequests = ref(false) const autoPauseOnExpired = ref(false) +const autoPause5hThreshold = ref(null) +const autoPause7dThreshold = ref(null) const mixedScheduling = ref(false) // For antigravity accounts: enable mixed scheduling const allowOverages = ref(false) // For antigravity accounts: enable AI Credits overages const antigravityModelRestrictionMode = ref<'whitelist' | 'mapping'>('whitelist') @@ -2862,9 +2896,11 @@ const syncFormFromAccount = (newAccount: Account | null) => { // Load mixed scheduling setting (only for antigravity accounts) mixedScheduling.value = false allowOverages.value = false - const extra = newAccount.extra as Record | undefined - mixedScheduling.value = extra?.mixed_scheduling === true - allowOverages.value = extra?.allow_overages === true + const extra = newAccount.extra as Record | undefined + mixedScheduling.value = extra?.mixed_scheduling === true + allowOverages.value = extra?.allow_overages === true + autoPause5hThreshold.value = typeof extra?.auto_pause_5h_threshold === 'number' ? extra.auto_pause_5h_threshold * 100 : null + autoPause7dThreshold.value = typeof extra?.auto_pause_7d_threshold === 'number' ? extra.auto_pause_7d_threshold * 100 : null // Load OpenAI passthrough toggle (OpenAI OAuth/API Key) openaiPassthroughEnabled.value = false @@ -3987,9 +4023,9 @@ const handleSubmit = async () => { } // For OpenAI OAuth/API Key accounts, handle passthrough mode in extra - if (props.account.platform === 'openai' && (props.account.type === 'oauth' || props.account.type === 'apikey')) { - const currentExtra = (props.account.extra as Record) || {} - const newExtra: Record = { ...currentExtra } + if (props.account.platform === 'openai' && (props.account.type === 'oauth' || props.account.type === 'apikey')) { + const currentExtra = (props.account.extra as Record) || {} + const newExtra: Record = { ...currentExtra } const hadCodexCLIOnlyEnabled = currentExtra.codex_cli_only === true if (props.account.type === 'oauth') { newExtra.openai_oauth_responses_websockets_v2_mode = openaiOAuthResponsesWebSocketV2Mode.value @@ -4011,15 +4047,25 @@ const handleSubmit = async () => { } else { newExtra.openai_compact_mode = openAICompactMode.value } - if (props.account.type === 'apikey') { + if (props.account.type === 'apikey') { if (!openAITextGenerationCapabilityEnabled.value || openAIResponsesMode.value === 'auto') { delete newExtra.openai_responses_mode } else { newExtra.openai_responses_mode = openAIResponsesMode.value } - } + } + if (autoPause5hThreshold.value != null && autoPause5hThreshold.value > 0) { + newExtra.auto_pause_5h_threshold = autoPause5hThreshold.value / 100 + } else { + delete newExtra.auto_pause_5h_threshold + } + if (autoPause7dThreshold.value != null && autoPause7dThreshold.value > 0) { + newExtra.auto_pause_7d_threshold = autoPause7dThreshold.value / 100 + } else { + delete newExtra.auto_pause_7d_threshold + } - delete newExtra.codex_image_generation_bridge_enabled + delete newExtra.codex_image_generation_bridge_enabled if (codexImageGenerationBridgeMode.value === 'inherit') { delete newExtra.codex_image_generation_bridge } else { diff --git a/frontend/src/components/account/__tests__/EditAccountModal.spec.ts b/frontend/src/components/account/__tests__/EditAccountModal.spec.ts index 4561924f..6db63831 100644 --- a/frontend/src/components/account/__tests__/EditAccountModal.spec.ts +++ b/frontend/src/components/account/__tests__/EditAccountModal.spec.ts @@ -330,6 +330,28 @@ describe('EditAccountModal', () => { ]) }) + it('submits OpenAI quota auto-pause thresholds in extra', async () => { + const account = buildAccount() + account.extra = { + auto_pause_5h_threshold: 0.9, + auto_pause_7d_threshold: 0.8 + } + updateAccountMock.mockReset() + checkMixedChannelRiskMock.mockReset() + checkMixedChannelRiskMock.mockResolvedValue({ has_risk: false }) + updateAccountMock.mockResolvedValue(account) + + const wrapper = mountModal(account) + + await wrapper.get('[data-testid="auto-pause-5h-threshold"]').setValue('95') + await wrapper.get('[data-testid="auto-pause-7d-threshold"]').setValue('96') + await wrapper.get('form#edit-account-form').trigger('submit.prevent') + + expect(updateAccountMock).toHaveBeenCalledTimes(1) + expect(updateAccountMock.mock.calls[0]?.[1]?.extra?.auto_pause_5h_threshold).toBe(0.95) + expect(updateAccountMock.mock.calls[0]?.[1]?.extra?.auto_pause_7d_threshold).toBe(0.96) + }) + it('keeps at least one OpenAI APIKey endpoint capability selected', async () => { const account = buildAccount() updateAccountMock.mockReset() diff --git a/frontend/src/i18n/locales/en.ts b/frontend/src/i18n/locales/en.ts index b2aeb2f8..fa2e5a92 100644 --- a/frontend/src/i18n/locales/en.ts +++ b/frontend/src/i18n/locales/en.ts @@ -3475,6 +3475,9 @@ export default { 'When enabled, warmup requests like title generation will return mock responses without consuming upstream tokens', autoPauseOnExpired: 'Auto Pause On Expired', autoPauseOnExpiredDesc: 'When enabled, the account will auto pause scheduling after it expires', + autoPause5hThreshold: '5h Usage Threshold (%)', + autoPause7dThreshold: '7d Usage Threshold (%)', + autoPauseThresholdHint: 'Leave empty or set 0 to disable. Reaching the threshold only skips the account during scheduling and does not modify schedulable.', // Quota control (Anthropic OAuth/SetupToken only) quotaControl: { title: 'Quota Control', diff --git a/frontend/src/i18n/locales/zh.ts b/frontend/src/i18n/locales/zh.ts index 85d1feee..2364f9c4 100644 --- a/frontend/src/i18n/locales/zh.ts +++ b/frontend/src/i18n/locales/zh.ts @@ -3613,6 +3613,9 @@ export default { interceptWarmupRequestsDesc: '启用后，标题生成等预热请求将返回 mock 响应，不消耗上游 token', autoPauseOnExpired: '过期自动暂停调度', autoPauseOnExpiredDesc: '启用后，账号过期将自动暂停调度', + autoPause5hThreshold: '5h 用量阈值(%)', + autoPause7dThreshold: '7d 用量阈值(%)', + autoPauseThresholdHint: '填 0 或留空表示不启用；达到阈值后仅在调度时跳过账号，不修改 schedulable。', // Quota control (Anthropic OAuth/SetupToken only) quotaControl: { title: '配额控制',