[FSTORE-1633] Fix engine choice in case of connection to serverless (4.1) (#427)

* [FSTORE-1633] Fix engine choice in case of connection to serverless

* Bump version to 4.1.3

Author: aversey

java/beam/pom.xml

@@ -5,7 +5,7 @@
   <parent>
     <artifactId>hsfs-parent</artifactId>
     <groupId>com.logicalclocks</groupId>
-    <version>4.1.2</version>
+    <version>4.1.3</version>
   </parent>
   <modelVersion>4.0.0</modelVersion>
 

java/flink/pom.xml

@@ -5,7 +5,7 @@
   <parent>
     <artifactId>hsfs-parent</artifactId>
     <groupId>com.logicalclocks</groupId>
-    <version>4.1.2</version>
+    <version>4.1.3</version>
   </parent>
   <modelVersion>4.0.0</modelVersion>
 

java/hsfs/pom.xml

@@ -5,7 +5,7 @@
   <parent>
     <artifactId>hsfs-parent</artifactId>
     <groupId>com.logicalclocks</groupId>
-    <version>4.1.2</version>
+    <version>4.1.3</version>
   </parent>
   <modelVersion>4.0.0</modelVersion>
 

java/pom.xml

@@ -7,7 +7,7 @@
     <groupId>com.logicalclocks</groupId>
     <artifactId>hsfs-parent</artifactId>
     <packaging>pom</packaging>
-    <version>4.1.2</version>
+    <version>4.1.3</version>
     <modules>
         <module>hsfs</module>
         <module>spark</module>

java/spark/pom.xml

@@ -22,7 +22,7 @@
     <parent>
         <artifactId>hsfs-parent</artifactId>
         <groupId>com.logicalclocks</groupId>
-        <version>4.1.2</version>
+        <version>4.1.3</version>
     </parent>
     <modelVersion>4.0.0</modelVersion>
 

python/hopsworks_common/connection.py

@@ -24,7 +24,7 @@
 import weakref
 from typing import Any, Optional
 
-from hopsworks_common import client, usage, util, version
+from hopsworks_common import client, constants, usage, util, version
 from hopsworks_common.core import (
     hosts_api,
     project_api,
@@ -98,13 +98,12 @@ class Connection:
         project: The name of the project to connect to. When running on Hopsworks, this
             defaults to the project from where the client is run from.
             Defaults to `None`.
-        engine: Which engine to use, `"spark"`, `"python"` or `"training"`. Defaults to `None`,
-            which initializes the engine to Spark if the environment provides Spark, for
-            example on Hopsworks and Databricks, or falls back to Python if Spark is not
-            available, e.g. on local Python environments or AWS SageMaker. This option
-            allows you to override this behaviour. `"training"` engine is useful when only
-            feature store metadata is needed, for example training dataset location and label
-            information when Hopsworks training experiment is conducted.
+        engine: Specifies the engine to use. Possible options are "spark", "python", "training", "spark-no-metastore", or "spark-delta". The default value is None, which automatically selects the engine based on the environment:
+            "spark": Used if Spark is available and the connection is not to serverless Hopsworks, such as in Hopsworks or Databricks environments.
+            "python": Used in local Python environments or AWS SageMaker when Spark is not available or the connection is done to serverless Hopsworks.
+            "training": Used when only feature store metadata is needed, such as for obtaining training dataset locations and label information during Hopsworks training experiments.
+            "spark-no-metastore": Functions like "spark" but does not rely on the Hive metastore.
+            "spark-delta": Minimizes dependencies further by avoiding both Hive metastore and HopsFS.
         hostname_verification: Whether or not to verify Hopsworks' certificate, defaults
             to `True`.
         trust_store_path: Path on the file system containing the Hopsworks certificates,
@@ -338,30 +337,35 @@ def connect(self) -> None:
         self._connected = True
         finalizer = weakref.finalize(self, self.close)
         try:
+            external = client.base.Client.REST_ENDPOINT not in os.environ
+            serverless = self._host == constants.HOSTS.APP_HOST
             # determine engine, needed to init client
-            if (self._engine is not None and self._engine.lower() == "spark") or (
-                self._engine is None and importlib.util.find_spec("pyspark")
+            if (
+                self._engine is None
+                and importlib.util.find_spec("pyspark")
+                and (not external or not serverless)
             ):
                 self._engine = "spark"
-            elif (self._engine is not None and self._engine.lower() == "python") or (
-                self._engine is None and not importlib.util.find_spec("pyspark")
-            ):
+            elif self._engine is None:
+                self._engine = "python"
+            elif self._engine.lower() == "spark":
+                self._engine = "spark"
+            elif self._engine.lower() == "python":
                 self._engine = "python"
-            elif self._engine is not None and self._engine.lower() == "training":
+            elif self._engine.lower() == "training":
                 self._engine = "training"
-            elif (
-                self._engine is not None
-                and self._engine.lower() == "spark-no-metastore"
-            ):
+            elif self._engine.lower() == "spark-no-metastore":
                 self._engine = "spark-no-metastore"
+            elif self._engine.lower() == "spark-delta":
+                self._engine = "spark-delta"
             else:
                 raise ConnectionError(
                     "Engine you are trying to initialize is unknown. "
                     "Supported engines are `'spark'`, `'python'` and `'training'`."
                 )
 
             # init client
-            if client.base.Client.REST_ENDPOINT not in os.environ:
+            if external:
                 client.init(
                     "external",
                     self._host,

python/hopsworks_common/version.py

@@ -14,4 +14,4 @@
 #   limitations under the License.
 #
 
-__version__ = "4.1.2"
+__version__ = "4.1.3"

utils/java/pom.xml

@@ -5,7 +5,7 @@
 
     <groupId>com.logicalclocks</groupId>
     <artifactId>hsfs-utils</artifactId>
-    <version>4.1.2</version>
+    <version>4.1.3</version>
 
     <properties>
         <hops.version>3.2.0.0-SNAPSHOT</hops.version>