fix(collections): handle errors from LocalRecall constructors

LocalRecall's NewPersistent{Chrome,LocalAI,Postgres}Collection now
return errors instead of os.Exit'ing on init failure. Bump the dep and
update newVectorEngine to consume the new signatures, surfacing failures
as a nil collection (the existing "engine misconfigured" code path)
instead of letting the upstream os.Exit kill the embedding process.

Without this, a transient embedding-service or PostgreSQL outage during
lazy RAG init would crash any long-running host (e.g. the LocalAI
distributed orchestrator) instead of degrading to "no RAG available".

RAGProviderFromState already handles a nil collection from
EnsureCollection by returning (nil, nil, false) to the agent pool, which
is the same path agents take when no RAG is configured — so callers
upstream need no further change.

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

Author: mudler

go.mod

@@ -18,7 +18,7 @@ require (
 	github.com/jung-kurt/gofpdf v1.16.2
 	github.com/modelcontextprotocol/go-sdk v1.2.0
 	github.com/mudler/cogito v0.9.5-0.20260315222927-63abdec7189b
-	github.com/mudler/localrecall v0.5.9-0.20260415164846-8ad831f840fc
+	github.com/mudler/localrecall v0.5.10-0.20260504162944-6138c1f535ab
 	github.com/mudler/skillserver v0.0.5-0.20260221145827-0639a82c8f49
 	github.com/mudler/xlog v0.0.5
 	github.com/onsi/ginkgo/v2 v2.27.5

go.sum

@@ -299,8 +299,8 @@ github.com/mschoch/smat v0.2.0 h1:8imxQsjDm8yFEAVBe7azKmKSgzSkZXDuKkSq9374khM=
 github.com/mschoch/smat v0.2.0/go.mod h1:kc9mz7DoBKqDyiRL7VZN8KvXQMWeTaVnttLRXOlotKw=
 github.com/mudler/cogito v0.9.5-0.20260315222927-63abdec7189b h1:A74T2Lauvg61KodYqsjTYDY05kPLcW+efVZjd23dghU=
 github.com/mudler/cogito v0.9.5-0.20260315222927-63abdec7189b/go.mod h1:6sfja3lcu2nWRzEc0wwqGNu/eCG3EWgij+8s7xyUeQ4=
-github.com/mudler/localrecall v0.5.9-0.20260415164846-8ad831f840fc h1:p1ucQ2rbU4mhG2Xl1Emg5Q6QCYCjI+fvMF9KTek/+sY=
-github.com/mudler/localrecall v0.5.9-0.20260415164846-8ad831f840fc/go.mod h1:xuPtgL9zUyiQLmspYzO3kaboYrGbWmwi8BQPt1aCAcs=
+github.com/mudler/localrecall v0.5.10-0.20260504162944-6138c1f535ab h1:U6MWVv9Xgb56JTIL4DfsZftSig/LeJA+yizlyw8fq24=
+github.com/mudler/localrecall v0.5.10-0.20260504162944-6138c1f535ab/go.mod h1:xuPtgL9zUyiQLmspYzO3kaboYrGbWmwi8BQPt1aCAcs=
 github.com/mudler/skillserver v0.0.5-0.20260221145827-0639a82c8f49 h1:dAF1ALXqqapRZo80x56BIBBcPrPbRNerbd66rdyO8J4=
 github.com/mudler/skillserver v0.0.5-0.20260221145827-0639a82c8f49/go.mod h1:z3yFhcL9bSykmmh6xgGu0hyoItd4CnxgtWMEWw8uFJU=
 github.com/mudler/xlog v0.0.5 h1:2unBuVC5rNGhCC86UaA94TElWFml80NL5XLK+kAmNuU=

webui/collections/inprocess.go

@@ -13,30 +13,45 @@ import (
 	"github.com/sashabaranov/go-openai"
 )
 
+// newVectorEngine creates the underlying RAG store for a collection.
+// Returns nil + logs on failure (engine misconfiguration or transient embedding/DB
+// outage). Callers must check for nil — RAG init failures must not crash the
+// embedding application; LocalRecall used to os.Exit on these errors but now
+// returns them, and we surface that to callers as a nil collection.
 func newVectorEngine(
 	vectorEngineType string,
 	llmClient *openai.Client,
 	apiURL, apiKey, collectionName, dbPath, fileAssets, embeddingModel, databaseURL string,
 	maxChunkSize, chunkOverlap int,
 ) *rag.PersistentKB {
+	var (
+		kb  *rag.PersistentKB
+		err error
+	)
 	switch vectorEngineType {
 	case "chromem":
 		xlog.Info("Chromem collection", "collectionName", collectionName, "dbPath", dbPath)
-		return rag.NewPersistentChromeCollection(llmClient, collectionName, dbPath, fileAssets, embeddingModel, maxChunkSize, chunkOverlap)
+		kb, err = rag.NewPersistentChromeCollection(llmClient, collectionName, dbPath, fileAssets, embeddingModel, maxChunkSize, chunkOverlap)
 	case "localai":
 		xlog.Info("LocalAI collection", "collectionName", collectionName, "apiURL", apiURL)
-		return rag.NewPersistentLocalAICollection(llmClient, apiURL, apiKey, collectionName, dbPath, fileAssets, embeddingModel, maxChunkSize, chunkOverlap)
+		kb, err = rag.NewPersistentLocalAICollection(llmClient, apiURL, apiKey, collectionName, dbPath, fileAssets, embeddingModel, maxChunkSize, chunkOverlap)
 	case "postgres":
 		if databaseURL == "" {
 			xlog.Error("DATABASE_URL is required for PostgreSQL engine")
 			return nil
 		}
 		xlog.Info("PostgreSQL collection", "collectionName", collectionName, "databaseURL", databaseURL)
-		return rag.NewPersistentPostgresCollection(llmClient, collectionName, dbPath, fileAssets, embeddingModel, maxChunkSize, chunkOverlap, databaseURL)
+		kb, err = rag.NewPersistentPostgresCollection(llmClient, collectionName, dbPath, fileAssets, embeddingModel, maxChunkSize, chunkOverlap, databaseURL)
 	default:
 		xlog.Error("Unknown vector engine", "engine", vectorEngineType)
 		return nil
 	}
+	if err != nil {
+		xlog.Error("Failed to create vector engine collection",
+			"engine", vectorEngineType, "collection", collectionName, "error", err)
+		return nil
+	}
+	return kb
 }
 
 // backendInProcess implements Backend using in-process state.