direct: Engine.SaveState takes ctx, returns void; logs I/O failures internally

Resource implementations cannot recover from a state-persistence failure —
the resource already exists on the server and aborting would not undo its
creation. Log the error via logdiag instead of propagating it, and drop the
error return so call sites are a single statement.

Co-authored-by: Denis Bilenko

Author: denik

bundle/direct/dresources/adapter.go

@@ -53,7 +53,7 @@ type IResource interface {
 
 	// DoCreate creates a new resource from the newState. Returns id of the resource and optionally remote state.
 	// If remote state is available as part of the operation, return it; otherwise return nil.
-	// Call engine.SaveState(id, state) to persist intermediate state before long-running waits.
+	// Call engine.SaveState(ctx, id, state) to persist intermediate state before long-running waits.
 	// Example: func (r *ResourceVolume) DoCreate(ctx context.Context, _ *Engine, newState *catalog.CreateVolumeRequestContent) (string, *catalog.VolumeInfo, error)
 	DoCreate(ctx context.Context, engine *Engine, newState any) (id string, remoteState any, e error)
 

bundle/direct/dresources/app.go

@@ -157,9 +157,7 @@ func (r *ResourceApp) DoCreate(ctx context.Context, engine *Engine, config *AppS
 	// Save state as soon as the app exists so it is not orphaned if the wait or
 	// lifecycle management is interrupted.
 
-	if err := engine.SaveState(app.Name, config); err != nil {
-		return "", nil, err
-	}
+	engine.SaveState(ctx, app.Name, config)
 
 	remote, err := r.waitForApp(ctx, r.client, config.Name)
 	if err != nil {

bundle/direct/dresources/cluster.go

@@ -156,9 +156,7 @@ func (r *ResourceCluster) DoCreate(ctx context.Context, engine *Engine, config *
 
 	// Save state immediately after the cluster is created so it is not orphaned
 	// if the subsequent wait or terminate is interrupted.
-	if err := engine.SaveState(id, config); err != nil {
-		return "", nil, err
-	}
+	engine.SaveState(ctx, id, config)
 
 	// Always wait for RUNNING first: clusters start in PENDING state and must be polled.
 	_, err = r.client.Clusters.WaitGetClusterRunning(ctx, id, 15*time.Minute, nil)

bundle/direct/dresources/database_instance.go

@@ -36,9 +36,7 @@ func (d *ResourceDatabaseInstance) DoCreate(ctx context.Context, engine *Engine,
 
 	// Save state immediately after the instance is created so it is not orphaned
 	// if the subsequent wait is interrupted.
-	if err := engine.SaveState(id, config); err != nil {
-		return "", nil, err
-	}
+	engine.SaveState(ctx, id, config)
 
 	waiterObj := &database.WaitGetDatabaseInstanceDatabaseAvailable[database.DatabaseInstance]{
 		Response: config,

bundle/direct/dresources/engine.go

@@ -1,8 +1,11 @@
 package dresources
 
 import (
+	"context"
 	"fmt"
 	"reflect"
+
+	"github.com/databricks/cli/libs/logdiag"
 )
 
 // Engine provides state persistence to resource implementations.
@@ -28,15 +31,19 @@ func NewNopEngine(stateType reflect.Type) *Engine {
 // SaveState saves the resource state. id must be the resource's identifier; on
 // the first call it is recorded, and subsequent calls panic if a different id is
 // passed. x must be a pointer to the same struct type as the resource's state.
-func (e *Engine) SaveState(id string, x any) error {
+// Failures to persist state are logged but do not abort the deployment — the
+// resource already exists and aborting would not undo its creation.
+func (e *Engine) SaveState(ctx context.Context, id string, x any) {
 	if e.id == "" {
 		e.id = id
 	} else if e.id != id {
 		panic(fmt.Sprintf("SaveState: id mismatch: expected %q, got %q", e.id, id))
 	}
 	xt := reflect.TypeOf(x)
 	if xt != e.stateType {
-		return fmt.Errorf("SaveState: type mismatch: expected %v, got %v", e.stateType, xt)
+		panic(fmt.Sprintf("SaveState: type mismatch: expected %v, got %v", e.stateType, xt))
+	}
+	if err := e.saveFunc(e.id, x); err != nil {
+		logdiag.LogError(ctx, err)
 	}
-	return e.saveFunc(e.id, x)
 }

bundle/direct/dresources/model_serving_endpoint.go

@@ -137,9 +137,7 @@ func (r *ResourceModelServingEndpoint) DoCreate(ctx context.Context, engine *Eng
 
 	// Save state immediately after the endpoint is created so it is not orphaned
 	// if the subsequent wait is interrupted.
-	if err := engine.SaveState(id, config); err != nil {
-		return "", nil, err
-	}
+	engine.SaveState(ctx, id, config)
 
 	remote, err := r.waitForEndpointReady(ctx, config.Name)
 	return id, remote, err

bundle/direct/dresources/vector_search_endpoint.go

@@ -74,9 +74,7 @@ func (r *ResourceVectorSearchEndpoint) DoCreate(ctx context.Context, engine *Eng
 
 	// Save state immediately after the endpoint is created so it is not orphaned
 	// if the subsequent wait is interrupted.
-	if err := engine.SaveState(id, config); err != nil {
-		return "", nil, err
-	}
+	engine.SaveState(ctx, id, config)
 
 	info, err := r.client.VectorSearchEndpoints.WaitGetEndpointVectorSearchEndpointOnline(ctx, config.Name, 60*time.Minute, nil)
 	if err != nil {

bundle/direct/dresources/vector_search_index.go

@@ -138,9 +138,7 @@ func (r *ResourceVectorSearchIndex) DoCreate(ctx context.Context, engine *Engine
 
 	// Save state immediately after the index is created (endpoint UUID now set) so it
 	// is not orphaned if the subsequent provisioning wait is interrupted.
-	if err := engine.SaveState(config.Name, config); err != nil {
-		return "", nil, err
-	}
+	engine.SaveState(ctx, config.Name, config)
 
 	remote, err := r.waitForIndexReady(ctx, config.Name, endpointUuid)
 	if err != nil {