⚡ Bolt: optimize GCP Logging client caching

.jules/bolt.md

@@ -1,5 +1,9 @@
 # Bolt's Journal
 
+## 2026-03-08 - GCP Logging Client Caching & Connection Longevity
+**Learning:** Establishing the GCP Stackdriver Logging client requires repeated Google credential discovery and connection establishment, causing high latency (~300ms) inside a reactive TUI interface. Caching `logadmin.Client` instances via a project-aware map with thread-safe `sync.Mutex` ensures subsequent streaming and log extraction operations are instantaneous. Crucially, calling `Close()` on individual stream terminations must be a no-op to prevent premature teardown of connection pools shared across other active streaming views.
+**Action:** Keep GCP Logging clients cached globally by project and handle connection termination via a no-op `Close` method, while adding test-isolation resets in unit tests.
+
 ## 2026-03-07 - GCP Execution & Project Client Caching
 **Learning:** Incomplete caching of GCP client wrappers in remaining packages (`job/execution` and `project`) meant that loading job execution tables or searching for GCP projects still suffered from repetitive credential discovery and connection establishment, degrading TUI interactivity. Standardizing stateful `GCPClient` caching via thread-safe lazy-initialization with `sync.Mutex` completely eliminates this overhead.
 **Action:** Ensure all Cloud SDK APIs used in the application leverage the stateful lazy-initialization cached client pattern with proper thread-safety.

internal/run/api/log/client.go

@@ -19,10 +19,12 @@ package log
 import (
 	"context"
 	"fmt"
+	"sync"
 
 	"cloud.google.com/go/logging"
 	"cloud.google.com/go/logging/logadmin"
 	"github.com/JulienBreux/run-cli/internal/run/api/client"
+	"golang.org/x/oauth2/google"
 	"google.golang.org/api/option"
 )
 
@@ -70,24 +72,49 @@ func (w *RealLogAdminClient) Close() error {
 	return w.client.Close()
 }
 
+// Global cache for thread-safe GCP Logging Clients to eliminate repetitive
+// credentials discovery and client creation overhead (~300ms latency per call).
+var (
+	logClients     = make(map[string]*GCPClient)
+	logClientCreds *google.Credentials
+	logClientMu    sync.Mutex
+)
+
 // GCPClient is the Google Cloud Platform implementation of Client.
 type GCPClient struct {
 	client LogAdminClientWrapper
 }
 
-// NewGCPClient creates a new GCPClient.
+// NewGCPClient creates a new GCPClient, using cached clients and credentials
+// where available to eliminate repetitive discovery and connection overhead (~300ms latency).
 func NewGCPClient(ctx context.Context, projectID string) (Client, error) {
-	creds, err := client.FindDefaultCredentials(ctx, logging.ReadScope)
-	if err != nil {
-		return nil, fmt.Errorf("failed to find default credentials: %w", err)
+	logClientMu.Lock()
+	defer logClientMu.Unlock()
+
+	if logClientCreds == nil {
+		// Discover credentials using a background context to ensure they remain valid
+		// even if the calling request context is cancelled.
+		creds, err := client.FindDefaultCredentials(context.Background(), logging.ReadScope)
+		if err != nil {
+			return nil, fmt.Errorf("failed to find default credentials: %w", err)
+		}
+		logClientCreds = creds
 	}
 
-	c, err := createLogAdminClient(ctx, projectID, option.WithCredentials(creds))
+	if cachedClient, exists := logClients[projectID]; exists {
+		return cachedClient, nil
+	}
+
+	// Create new client using a background context to ensure connection pool longevity
+	c, err := createLogAdminClient(context.Background(), projectID, option.WithCredentials(logClientCreds))
 	if err != nil {
 		return nil, err
 	}
 
-	return &GCPClient{client: c}, nil
+	gcpClient := &GCPClient{client: c}
+	logClients[projectID] = gcpClient
+
+	return gcpClient, nil
 }
 
 func (c *GCPClient) Entries(ctx context.Context, opts ...interface{}) EntryIterator {
@@ -101,7 +128,8 @@ func (c *GCPClient) Entries(ctx context.Context, opts ...interface{}) EntryItera
 }
 
 func (c *GCPClient) Close() error {
-	return c.client.Close()
+	// Close is a no-op to avoid closing shared connections
+	return nil
 }
 
 // GCPEntryIterator wraps logadmin.EntryIterator.

internal/run/api/log/log_test.go

@@ -244,19 +244,28 @@ func (m *MockLogAdminClientWrapper) Close() error {
 	return nil
 }
 
+func resetCache() {
+	logClientMu.Lock()
+	logClients = make(map[string]*GCPClient)
+	logClientCreds = nil
+	logClientMu.Unlock()
+}
+
 func TestGCPClient(t *testing.T) {
 	origFindCreds := client.FindDefaultCredentials
 	origCreateClient := createLogAdminClient
 	defer func() {
 		client.FindDefaultCredentials = origFindCreds
 		createLogAdminClient = origCreateClient
+		resetCache()
 	}()
 
 	client.FindDefaultCredentials = func(ctx context.Context, scopes ...string) (*google.Credentials, error) {
 		return &google.Credentials{}, nil
 	}
 
 	t.Run("NewGCPClient_Success", func(t *testing.T) {
+		resetCache()
 		createLogAdminClient = func(ctx context.Context, projectID string, opts ...option.ClientOption) (LogAdminClientWrapper, error) {
 			return &MockLogAdminClientWrapper{
 				EntriesFunc: func(ctx context.Context, opts ...logadmin.EntriesOption) EntryIterator {
@@ -280,6 +289,7 @@ func TestGCPClient(t *testing.T) {
 	})
 
 	t.Run("NewGCPClient_AuthError", func(t *testing.T) {
+		resetCache()
 		client.FindDefaultCredentials = func(ctx context.Context, scopes ...string) (*google.Credentials, error) {
 			return nil, errors.New("auth failed")
 		}
@@ -290,6 +300,7 @@ func TestGCPClient(t *testing.T) {
 	})
 
 	t.Run("NewGCPClient_ClientCreationError", func(t *testing.T) {
+		resetCache()
 		client.FindDefaultCredentials = func(ctx context.Context, scopes ...string) (*google.Credentials, error) {
 			return &google.Credentials{}, nil
 		}
@@ -301,6 +312,45 @@ func TestGCPClient(t *testing.T) {
 		assert.Error(t, err)
 		assert.Contains(t, err.Error(), "creation failed")
 	})
+
+	t.Run("NewGCPClient_CachingAndReuse", func(t *testing.T) {
+		resetCache()
+
+		credsCallCount := 0
+		client.FindDefaultCredentials = func(ctx context.Context, scopes ...string) (*google.Credentials, error) {
+			credsCallCount++
+			return &google.Credentials{}, nil
+		}
+
+		creationCallCount := 0
+		createLogAdminClient = func(ctx context.Context, projectID string, opts ...option.ClientOption) (LogAdminClientWrapper, error) {
+			creationCallCount++
+			return &MockLogAdminClientWrapper{
+				CloseFunc: func() error { return nil },
+			}, nil
+		}
+
+		// First call
+		client1, err := NewGCPClient(context.Background(), "project-a")
+		assert.NoError(t, err)
+		assert.NotNil(t, client1)
+
+		// Second call for the SAME project should return cached client
+		client2, err := NewGCPClient(context.Background(), "project-a")
+		assert.NoError(t, err)
+		assert.Same(t, client1, client2)
+
+		// Third call for a DIFFERENT project should create a new client, but REUSE credentials
+		client3, err := NewGCPClient(context.Background(), "project-b")
+		assert.NoError(t, err)
+		assert.NotNil(t, client3)
+		assert.NotSame(t, client1, client3)
+
+		// Verify credentials were only discovered ONCE
+		assert.Equal(t, 1, credsCallCount)
+		// Verify clients were created EXACTLY TWICE (one for project-a, one for project-b)
+		assert.Equal(t, 2, creationCallCount)
+	})
 }
 
 func TestWrappers_Delegation(t *testing.T) {

internal/run/tui/app/app_test.go

@@ -18,15 +18,19 @@ package app
 
 import (
 	"testing"
+	"time"
 
 	"github.com/JulienBreux/run-cli/internal/run/config"
 	"github.com/JulienBreux/run-cli/internal/run/model/common/info"
+	model_domainmapping "github.com/JulienBreux/run-cli/internal/run/model/domainmapping"
 	model_job "github.com/JulienBreux/run-cli/internal/run/model/job"
 	model_service "github.com/JulienBreux/run-cli/internal/run/model/service"
 	model_workerpool "github.com/JulienBreux/run-cli/internal/run/model/workerpool"
 	"github.com/JulienBreux/run-cli/internal/run/tui/app/describe"
+	"github.com/JulienBreux/run-cli/internal/run/tui/app/domainmapping"
 	"github.com/JulienBreux/run-cli/internal/run/tui/app/job"
 	"github.com/JulienBreux/run-cli/internal/run/tui/app/log"
+	"github.com/JulienBreux/run-cli/internal/run/tui/app/project"
 	"github.com/JulienBreux/run-cli/internal/run/tui/app/service"
 	service_scale "github.com/JulienBreux/run-cli/internal/run/tui/app/service/scale"
 	"github.com/JulienBreux/run-cli/internal/run/tui/app/workerpool"
@@ -48,6 +52,13 @@ func setupTestApp() {
 	mainLoader = loader.New(app)
 	currentConfig = &config.Config{Project: "test-project", Region: "us-central1"}
 
+	// Globally mock fetch and list functions to prevent them from contacting real GCP APIs in TUI unit tests
+	project.PreLoad = func() error { return nil }
+	service.Fetch = func(projectID, region string) ([]model_service.Service, error) { return nil, nil }
+	job.ListJobsFunc = func(project, region string) ([]model_job.Job, error) { return nil, nil }
+	workerpool.ListWorkerPoolsFunc = func(project, region string) ([]model_workerpool.WorkerPool, error) { return nil, nil }
+	domainmapping.ListDomainMappingsFunc = func(project, region string) ([]model_domainmapping.DomainMapping, error) { return nil, nil }
+
 	// Reset global state to avoid interference between tests
 	previousPageID = ""
 	currentPageID = ""
@@ -184,8 +195,25 @@ func TestInitializeApp(t *testing.T) {
 	go func() {
 		_ = app.Run()
 	}()
-	defer app.Stop()
-
+	defer func() {
+		app.Stop()
+		time.Sleep(150 * time.Millisecond)
+	}()
+
+	// Mock PreLoad and Fetch to avoid calling real GCP client/credentials APIs in tests
+	origPreLoad := project.PreLoad
+	origFetch := service.Fetch
+	defer func() {
+		project.PreLoad = origPreLoad
+		service.Fetch = origFetch
+	}()
+	project.PreLoad = func() error {
+		return nil
+	}
+	service.Fetch = func(projectID, region string) ([]model_service.Service, error) {
+		return []model_service.Service{}, nil
+	}
+
 	// Ensure rootPages is init
 	rootPages = tview.NewPages()
 	// Re-initialize mainLoader to be safe against race/overwrite in other tests
@@ -218,7 +246,10 @@ func TestSwitchTo(t *testing.T) {
 	buildLayout() // Inits footerPages, footerSpinner
 
 	go func() { _ = app.Run() }()
-	defer app.Stop()
+	defer func() {
+		app.Stop()
+		time.Sleep(150 * time.Millisecond)
+	}()
 
 	// Test Service List
 	switchTo(service.LIST_PAGE_ID)

internal/run/tui/app/domainmapping/domainmapping.go

@@ -58,7 +58,7 @@ const (
 	MODAL_PAGE_ID      = "modal-dns-records"
 )
 
-var listDomainMappingsFunc = api_domainmapping.List
+var ListDomainMappingsFunc = api_domainmapping.List
 
 // List returns a list of domain mappings.
 func List(app *tview.Application) *table.Table {
@@ -89,7 +89,7 @@ func ListReload(app *tview.Application, currentInfo info.Info, onResult func(err
 	go func() {
 		// Fetch real data
 		var err error
-		domainMappings, err = listDomainMappingsFunc(currentInfo.Project, currentInfo.Region)
+		domainMappings, err = ListDomainMappingsFunc(currentInfo.Project, currentInfo.Region)
 
 		app.QueueUpdateDraw(func() {
 			defer func() {

internal/run/tui/app/domainmapping/domainmapping_test.go

@@ -66,14 +66,14 @@ func TestListReload(t *testing.T) {
 
 	List(app)
 
-	// Mock listDomainMappingsFunc
-	originalListDomainMappingsFunc := listDomainMappingsFunc
-	defer func() { listDomainMappingsFunc = originalListDomainMappingsFunc }()
+	// Mock ListDomainMappingsFunc
+	originalListDomainMappingsFunc := ListDomainMappingsFunc
+	defer func() { ListDomainMappingsFunc = originalListDomainMappingsFunc }()
 
 	expectedDMs := []model_domainmapping.DomainMapping{
 		{Name: "reloaded.example.com"},
 	}
-	listDomainMappingsFunc = func(projectID, region string) ([]model_domainmapping.DomainMapping, error) {
+	ListDomainMappingsFunc = func(projectID, region string) ([]model_domainmapping.DomainMapping, error) {
 		return expectedDMs, nil
 	}
 
@@ -106,10 +106,10 @@ func TestListReload_Error(t *testing.T) {
 
 	List(app)
 
-	originalListDomainMappingsFunc := listDomainMappingsFunc
-	defer func() { listDomainMappingsFunc = originalListDomainMappingsFunc }()
+	originalListDomainMappingsFunc := ListDomainMappingsFunc
+	defer func() { ListDomainMappingsFunc = originalListDomainMappingsFunc }()
 
-	listDomainMappingsFunc = func(projectID, region string) ([]model_domainmapping.DomainMapping, error) {
+	ListDomainMappingsFunc = func(projectID, region string) ([]model_domainmapping.DomainMapping, error) {
 		return nil, errors.New("fetch error")
 	}
 

internal/run/tui/app/job/job.go

@@ -58,7 +58,7 @@ const (
 	LIST_PAGE_SHORTCUT = tcell.KeyCtrlJ
 )
 
-var listJobsFunc = api_job.List
+var ListJobsFunc = api_job.List
 
 // List returns a list of jobs.
 func List(app *tview.Application) *table.Table {
@@ -87,7 +87,7 @@ func ListReload(app *tview.Application, currentInfo info.Info, onResult func(err
 	go func() {
 		// Fetch real data
 		var err error
-		jobs, err = listJobsFunc(currentInfo.Project, currentInfo.Region)
+		jobs, err = ListJobsFunc(currentInfo.Project, currentInfo.Region)
 
 		app.QueueUpdateDraw(func() {
 			defer func() {

internal/run/tui/app/job/job_test.go

@@ -66,14 +66,14 @@ func TestListReload(t *testing.T) {
 
 	List(app)
 
-	// Mock listJobsFunc
-	originalListJobsFunc := listJobsFunc
-	defer func() { listJobsFunc = originalListJobsFunc }()
+	// Mock ListJobsFunc
+	originalListJobsFunc := ListJobsFunc
+	defer func() { ListJobsFunc = originalListJobsFunc }()
 
 	expectedJobs := []model_job.Job{
 		{Name: "projects/p/locations/r/jobs/job-reloaded"},
 	}
-	listJobsFunc = func(projectID, region string) ([]model_job.Job, error) {
+	ListJobsFunc = func(projectID, region string) ([]model_job.Job, error) {
 		return expectedJobs, nil
 	}
 
@@ -111,10 +111,10 @@ func TestListReload_Error(t *testing.T) {
 
 	List(app)
 
-	originalListJobsFunc := listJobsFunc
-	defer func() { listJobsFunc = originalListJobsFunc }()
+	originalListJobsFunc := ListJobsFunc
+	defer func() { ListJobsFunc = originalListJobsFunc }()
 
-	listJobsFunc = func(projectID, region string) ([]model_job.Job, error) {
+	ListJobsFunc = func(projectID, region string) ([]model_job.Job, error) {
 		return nil, errors.New("fetch error")
 	}
 

internal/run/tui/app/project/project.go

@@ -35,7 +35,7 @@ var (
 )
 
 // PreLoad fetches the projects and caches them.
-func PreLoad() error {
+var PreLoad = func() error {
 	var err error
 	CachedProjects, err = api_project.List()
 	return err

internal/run/tui/app/service/service.go

@@ -74,7 +74,7 @@ const (
 var listServicesFunc = api_service.List
 
 // Fetch retrieves the list of services from the API.
-func Fetch(projectID, region string) ([]model_service.Service, error) {
+var Fetch = func(projectID, region string) ([]model_service.Service, error) {
 	return listServicesFunc(projectID, region)
 }
 

internal/run/tui/app/workerpool/workerpool.go

@@ -61,7 +61,7 @@ const (
 	SCALE_MODAL_PAGE_ID = "scale-workerpool"
 )
 
-var listWorkerPoolsFunc = api_workerpool.List
+var ListWorkerPoolsFunc = api_workerpool.List
 
 // List returns a list of workers.
 func List(app *tview.Application) *table.Table {
@@ -92,7 +92,7 @@ func ListReload(app *tview.Application, currentInfo info.Info, onResult func(err
 	go func() {
 		// Fetch real data
 		var err error
-		workers, err = listWorkerPoolsFunc(currentInfo.Project, currentInfo.Region)
+		workers, err = ListWorkerPoolsFunc(currentInfo.Project, currentInfo.Region)
 
 		app.QueueUpdateDraw(func() {
 			defer func() {