Merge pull request router-for-me#2576 from zilianpn/fix/disable-cooling-auth-errors

fix(auth): honor disable-cooling and enrich no-auth errors

Author: luispater

config.example.yaml

@@ -87,6 +87,9 @@ max-retry-credentials: 0
 # Maximum wait time in seconds for a cooled-down credential before triggering a retry.
 max-retry-interval: 30
 
+# When true, disable auth/model cooldown scheduling globally (prevents blackout windows after failure states).
+disable-cooling: false
+
 # Quota exceeded behavior
 quota-exceeded:
   switch-project: true # Whether to automatically switch to another project when a quota is exceeded

sdk/api/handlers/handlers.go

@@ -6,6 +6,7 @@ package handlers
 import (
 	"bytes"
 	"encoding/json"
+	"errors"
 	"fmt"
 	"net/http"
 	"strings"
@@ -492,6 +493,7 @@ func (h *BaseAPIHandler) ExecuteWithAuthManager(ctx context.Context, handlerType
 	opts.Metadata = reqMeta
 	resp, err := h.AuthManager.Execute(ctx, providers, req, opts)
 	if err != nil {
+		err = enrichAuthSelectionError(err, providers, normalizedModel)
 		status := http.StatusInternalServerError
 		if se, ok := err.(interface{ StatusCode() int }); ok && se != nil {
 			if code := se.StatusCode(); code > 0 {
@@ -538,6 +540,7 @@ func (h *BaseAPIHandler) ExecuteCountWithAuthManager(ctx context.Context, handle
 	opts.Metadata = reqMeta
 	resp, err := h.AuthManager.ExecuteCount(ctx, providers, req, opts)
 	if err != nil {
+		err = enrichAuthSelectionError(err, providers, normalizedModel)
 		status := http.StatusInternalServerError
 		if se, ok := err.(interface{ StatusCode() int }); ok && se != nil {
 			if code := se.StatusCode(); code > 0 {
@@ -588,6 +591,7 @@ func (h *BaseAPIHandler) ExecuteStreamWithAuthManager(ctx context.Context, handl
 	opts.Metadata = reqMeta
 	streamResult, err := h.AuthManager.ExecuteStream(ctx, providers, req, opts)
 	if err != nil {
+		err = enrichAuthSelectionError(err, providers, normalizedModel)
 		errChan := make(chan *interfaces.ErrorMessage, 1)
 		status := http.StatusInternalServerError
 		if se, ok := err.(interface{ StatusCode() int }); ok && se != nil {
@@ -697,7 +701,7 @@ func (h *BaseAPIHandler) ExecuteStreamWithAuthManager(ctx context.Context, handl
 								chunks = retryResult.Chunks
 								continue outer
 							}
-							streamErr = retryErr
+							streamErr = enrichAuthSelectionError(retryErr, providers, normalizedModel)
 						}
 					}
 
@@ -840,6 +844,54 @@ func replaceHeader(dst http.Header, src http.Header) {
 	}
 }
 
+func enrichAuthSelectionError(err error, providers []string, model string) error {
+	if err == nil {
+		return nil
+	}
+
+	var authErr *coreauth.Error
+	if !errors.As(err, &authErr) || authErr == nil {
+		return err
+	}
+
+	code := strings.TrimSpace(authErr.Code)
+	if code != "auth_not_found" && code != "auth_unavailable" {
+		return err
+	}
+
+	providerText := strings.Join(providers, ",")
+	if providerText == "" {
+		providerText = "unknown"
+	}
+	modelText := strings.TrimSpace(model)
+	if modelText == "" {
+		modelText = "unknown"
+	}
+
+	baseMessage := strings.TrimSpace(authErr.Message)
+	if baseMessage == "" {
+		baseMessage = "no auth available"
+	}
+	detail := fmt.Sprintf("%s (providers=%s, model=%s)", baseMessage, providerText, modelText)
+
+	// Clarify the most common alias confusion between Anthropic route names and internal provider keys.
+	if strings.Contains(","+providerText+",", ",claude,") {
+		detail += "; check Claude auth/key session and cooldown state via /v0/management/auth-files"
+	}
+
+	status := authErr.HTTPStatus
+	if status <= 0 {
+		status = http.StatusServiceUnavailable
+	}
+
+	return &coreauth.Error{
+		Code:       authErr.Code,
+		Message:    detail,
+		Retryable:  authErr.Retryable,
+		HTTPStatus: status,
+	}
+}
+
 // WriteErrorResponse writes an error message to the response writer using the HTTP status embedded in the message.
 func (h *BaseAPIHandler) WriteErrorResponse(c *gin.Context, msg *interfaces.ErrorMessage) {
 	status := http.StatusInternalServerError

sdk/api/handlers/handlers_error_response_test.go

@@ -5,10 +5,12 @@ import (
 	"net/http"
 	"net/http/httptest"
 	"reflect"
+	"strings"
 	"testing"
 
 	"github.com/gin-gonic/gin"
 	"github.com/router-for-me/CLIProxyAPI/v6/internal/interfaces"
+	coreauth "github.com/router-for-me/CLIProxyAPI/v6/sdk/cliproxy/auth"
 	sdkconfig "github.com/router-for-me/CLIProxyAPI/v6/sdk/config"
 )
 
@@ -66,3 +68,46 @@ func TestWriteErrorResponse_AddonHeadersEnabled(t *testing.T) {
 		t.Fatalf("X-Request-Id = %#v, want %#v", got, []string{"new-1", "new-2"})
 	}
 }
+
+func TestEnrichAuthSelectionError_DefaultsTo503WithContext(t *testing.T) {
+	in := &coreauth.Error{Code: "auth_not_found", Message: "no auth available"}
+	out := enrichAuthSelectionError(in, []string{"claude"}, "claude-sonnet-4-6")
+
+	var got *coreauth.Error
+	if !errors.As(out, &got) || got == nil {
+		t.Fatalf("expected coreauth.Error, got %T", out)
+	}
+	if got.StatusCode() != http.StatusServiceUnavailable {
+		t.Fatalf("status = %d, want %d", got.StatusCode(), http.StatusServiceUnavailable)
+	}
+	if !strings.Contains(got.Message, "providers=claude") {
+		t.Fatalf("message missing provider context: %q", got.Message)
+	}
+	if !strings.Contains(got.Message, "model=claude-sonnet-4-6") {
+		t.Fatalf("message missing model context: %q", got.Message)
+	}
+	if !strings.Contains(got.Message, "/v0/management/auth-files") {
+		t.Fatalf("message missing management hint: %q", got.Message)
+	}
+}
+
+func TestEnrichAuthSelectionError_PreservesExplicitStatus(t *testing.T) {
+	in := &coreauth.Error{Code: "auth_unavailable", Message: "no auth available", HTTPStatus: http.StatusTooManyRequests}
+	out := enrichAuthSelectionError(in, []string{"gemini"}, "gemini-2.5-pro")
+
+	var got *coreauth.Error
+	if !errors.As(out, &got) || got == nil {
+		t.Fatalf("expected coreauth.Error, got %T", out)
+	}
+	if got.StatusCode() != http.StatusTooManyRequests {
+		t.Fatalf("status = %d, want %d", got.StatusCode(), http.StatusTooManyRequests)
+	}
+}
+
+func TestEnrichAuthSelectionError_IgnoresOtherErrors(t *testing.T) {
+	in := errors.New("boom")
+	out := enrichAuthSelectionError(in, []string{"claude"}, "claude-sonnet-4-6")
+	if out != in {
+		t.Fatalf("expected original error to be returned unchanged")
+	}
+}

sdk/api/handlers/handlers_stream_bootstrap_test.go

@@ -2,10 +2,13 @@ package handlers
 
 import (
 	"context"
+	"errors"
 	"net/http"
+	"strings"
 	"sync"
 	"testing"
 
+	"github.com/router-for-me/CLIProxyAPI/v6/internal/interfaces"
 	"github.com/router-for-me/CLIProxyAPI/v6/internal/registry"
 	coreauth "github.com/router-for-me/CLIProxyAPI/v6/sdk/cliproxy/auth"
 	coreexecutor "github.com/router-for-me/CLIProxyAPI/v6/sdk/cliproxy/executor"
@@ -463,6 +466,76 @@ func TestExecuteStreamWithAuthManager_DoesNotRetryAfterFirstByte(t *testing.T) {
 	}
 }
 
+func TestExecuteStreamWithAuthManager_EnrichesBootstrapRetryAuthUnavailableError(t *testing.T) {
+	executor := &failOnceStreamExecutor{}
+	manager := coreauth.NewManager(nil, nil, nil)
+	manager.RegisterExecutor(executor)
+
+	auth1 := &coreauth.Auth{
+		ID:       "auth1",
+		Provider: "codex",
+		Status:   coreauth.StatusActive,
+		Metadata: map[string]any{"email": "test1@example.com"},
+	}
+	if _, err := manager.Register(context.Background(), auth1); err != nil {
+		t.Fatalf("manager.Register(auth1): %v", err)
+	}
+
+	registry.GetGlobalRegistry().RegisterClient(auth1.ID, auth1.Provider, []*registry.ModelInfo{{ID: "test-model"}})
+	t.Cleanup(func() {
+		registry.GetGlobalRegistry().UnregisterClient(auth1.ID)
+	})
+
+	handler := NewBaseAPIHandlers(&sdkconfig.SDKConfig{
+		Streaming: sdkconfig.StreamingConfig{
+			BootstrapRetries: 1,
+		},
+	}, manager)
+	dataChan, _, errChan := handler.ExecuteStreamWithAuthManager(context.Background(), "openai", "test-model", []byte(`{"model":"test-model"}`), "")
+	if dataChan == nil || errChan == nil {
+		t.Fatalf("expected non-nil channels")
+	}
+
+	var got []byte
+	for chunk := range dataChan {
+		got = append(got, chunk...)
+	}
+	if len(got) != 0 {
+		t.Fatalf("expected empty payload, got %q", string(got))
+	}
+
+	var gotErr *interfaces.ErrorMessage
+	for msg := range errChan {
+		if msg != nil {
+			gotErr = msg
+		}
+	}
+	if gotErr == nil {
+		t.Fatalf("expected terminal error")
+	}
+	if gotErr.StatusCode != http.StatusServiceUnavailable {
+		t.Fatalf("status = %d, want %d", gotErr.StatusCode, http.StatusServiceUnavailable)
+	}
+
+	var authErr *coreauth.Error
+	if !errors.As(gotErr.Error, &authErr) || authErr == nil {
+		t.Fatalf("expected coreauth.Error, got %T", gotErr.Error)
+	}
+	if authErr.Code != "auth_unavailable" {
+		t.Fatalf("code = %q, want %q", authErr.Code, "auth_unavailable")
+	}
+	if !strings.Contains(authErr.Message, "providers=codex") {
+		t.Fatalf("message missing provider context: %q", authErr.Message)
+	}
+	if !strings.Contains(authErr.Message, "model=test-model") {
+		t.Fatalf("message missing model context: %q", authErr.Message)
+	}
+
+	if executor.Calls() != 1 {
+		t.Fatalf("expected exactly one upstream call before retry path selection failure, got %d", executor.Calls())
+	}
+}
+
 func TestExecuteStreamWithAuthManager_PinnedAuthKeepsSameUpstream(t *testing.T) {
 	executor := &authAwareStreamExecutor{}
 	manager := coreauth.NewManager(nil, nil, nil)