fix: Use a general string to represent the transformation service endpoint.

shuchu · shuchu · commit b65a6a4defd5 · 2024-11-20T23:19:03.000-05:00
Signed-off-by: Shuchu Han &lt;shuchu.han@gmail.com&gt;
diff --git a/go/internal/feast/featurestore.go b/go/internal/feast/featurestore.go
@@ -3,9 +3,7 @@ package feast
 import (
 	"context"
 	"errors"
-	"fmt"
 	"os"
-	"strings"
 
 	"github.com/apache/arrow/go/v17/arrow/memory"
 	//"gopkg.in/DataDog/dd-trace-go.v1/ddtrace/tracer"
@@ -62,10 +60,12 @@ func NewFeatureStore(config *registry.RepoConfig, callback transformation.Transf
 	if err != nil {
 		return nil, err
 	}
-	sanitizedProjectName := strings.Replace(config.Project, "_", "-", -1)
-	productName := os.Getenv("PRODUCT")
-	endpoint := fmt.Sprintf("%s-transformations.%s.svc.cluster.local:80", sanitizedProjectName, productName)
-	transformationService, _ := transformation.NewGrpcTransformationService(config, endpoint)
+
+	// Use a scalable transformation service like Python Transformation Service as the Python version of transformation service is
+	// better for data calculation than the Go version
+	transformationServerEndpoint := os.Getenv("TRANSFORM_SERVER_ENDPOINT")
+
+	transformationService, _ := transformation.NewGrpcTransformationService(config, transformationServerEndpoint)
 
 	return &FeatureStore{
 		config:                 config,
diff --git a/go/internal/feast/registry/registry.go b/go/internal/feast/registry/registry.go
@@ -15,10 +15,10 @@ import (
 
 var REGISTRY_SCHEMA_VERSION string = "1"
 var REGISTRY_STORE_CLASS_FOR_SCHEME map[string]string = map[string]string{
-	"gs":    "GCSRegistryStore",
-	"s3":    "S3RegistryStore",
-	"file":  "FileRegistryStore",
-	"":      "FileRegistryStore",
+	"gs":   "GCSRegistryStore",
+	"s3":   "S3RegistryStore",
+	"file": "FileRegistryStore",
+	"":     "FileRegistryStore",
 }
 
 /*
diff --git a/go/internal/feast/server/server_commons.go b/go/internal/feast/server/server_commons.go
@@ -28,4 +28,4 @@ func LogWithSpanContext(span tracer.Span) zerolog.Logger {
 
 	return logger
 }
-	*/
+*/
diff --git a/go/internal/feast/transformation/transformation.go b/go/internal/feast/transformation/transformation.go
@@ -22,6 +22,8 @@ TransformationCallback is a Python callback function's expected signature.
 The function should accept name of the on demand feature view and pointers to input & output record batches.
 Each record batch is being passed as two pointers: pointer to array (data) and pointer to schema.
 Python function is expected to return number of rows added to the output record batch.
+
+[11-20-2024] Use a Transformation GRPC service, like the Python version one, for better scalability.
 */
 type TransformationCallback func(ODFVName string, inputArrPtr, inputSchemaPtr, outArrPtr, outSchemaPtr uintptr, fullFeatureNames bool) int
 
diff --git a/go/main.go b/go/main.go
@@ -18,7 +18,6 @@ import (
 	"google.golang.org/grpc"
 	"google.golang.org/grpc/health"
 	"google.golang.org/grpc/health/grpc_health_v1"
-
 	//grpctrace "gopkg.in/DataDog/dd-trace-go.v1/contrib/google.golang.org/grpc"
 	//"gopkg.in/DataDog/dd-trace-go.v1/ddtrace/tracer"
 )

Original file line number	Diff line number	Diff line change
`@@ -28,4 +28,4 @@ func LogWithSpanContext(span tracer.Span) zerolog.Logger {`
`28`	`28`
`29`	`29`	`return logger`
`30`	`30`	`}`
`31`		`- */`
	`31`	`+*/`
Original file line number	Diff line number	Diff line change
`@@ -18,7 +18,6 @@ import (`
`18`	`18`	`"google.golang.org/grpc"`
`19`	`19`	`"google.golang.org/grpc/health"`
`20`	`20`	`"google.golang.org/grpc/health/grpc_health_v1"`
`21`		`-`
`22`	`21`	`//grpctrace "gopkg.in/DataDog/dd-trace-go.v1/contrib/google.golang.org/grpc"`
`23`	`22`	`//"gopkg.in/DataDog/dd-trace-go.v1/ddtrace/tracer"`
`24`	`23`	`)`