diff --git a/.github/workflows/build_python_connect.yml b/.github/workflows/build_python_connect.yml
index bf247db613dba..cfcd6a2a5ec89 100644
--- a/.github/workflows/build_python_connect.yml
+++ b/.github/workflows/build_python_connect.yml
@@ -79,6 +79,8 @@ jobs:
         env:
           SPARK_TESTING: 1
           SPARK_CONNECT_TESTING_REMOTE: sc://localhost
+          # Increase socket timeout for CI environment reliability
+          SPARK_AUTH_SOCKET_TIMEOUT: 60
         run: |
           # Make less noisy
           cp conf/log4j2.properties.template conf/log4j2.properties
diff --git a/.github/workflows/build_python_connect35.yml b/.github/workflows/build_python_connect35.yml
index 6c37091afcb4f..6c0bf433505ba 100644
--- a/.github/workflows/build_python_connect35.yml
+++ b/.github/workflows/build_python_connect35.yml
@@ -22,6 +22,8 @@ name: Build / Python-only, Connect-only (master-server, branch-3.5-client, Pytho
 on:
   schedule:
     - cron: '0 21 * * *'
+  pull_request:
+    branches: [ master, main ]
   workflow_dispatch:
 
 jobs:
@@ -82,6 +84,8 @@ jobs:
           SPARK_TESTING: 1
           SPARK_SKIP_CONNECT_COMPAT_TESTS: 1
           SPARK_CONNECT_TESTING_REMOTE: sc://localhost
+          # Increase socket timeout for CI environment reliability
+          SPARK_AUTH_SOCKET_TIMEOUT: 60
         run: |
           # Make less noisy
           cp conf/log4j2.properties.template conf/log4j2.properties
diff --git a/python/pyspark/sql/connect/streaming/worker/foreach_batch_worker.py b/python/pyspark/sql/connect/streaming/worker/foreach_batch_worker.py
index b819634adb5a6..13aafe17a1c2e 100644
--- a/python/pyspark/sql/connect/streaming/worker/foreach_batch_worker.py
+++ b/python/pyspark/sql/connect/streaming/worker/foreach_batch_worker.py
@@ -55,6 +55,28 @@ def process(df_id, batch_id):  # type: ignore[no-untyped-def]
             f"{log_name} Completed batch {batch_id} with DF id {df_id} and session id {session_id}"
         )
 
+    def create_spark_session_with_retry(connect_url, session_id, max_retries=3):
+        """Create Spark Connect session with retry logic for better reliability in CI environments."""
+        import time
+        
+        for attempt in range(max_retries):
+            try:
+                print(f"{log_name} Attempting to connect (attempt {attempt + 1}/{max_retries})")
+                spark_connect_session = SparkSession.builder.remote(connect_url).getOrCreate()
+                assert spark_connect_session.session_id == session_id
+                print(f"{log_name} Successfully connected to Spark Connect server")
+                return spark_connect_session
+            except Exception as e:
+                print(f"{log_name} Connection attempt {attempt + 1} failed: {str(e)}")
+                if attempt < max_retries - 1:
+                    # Exponential backoff: 1s, 2s, 4s
+                    wait_time = 2 ** attempt
+                    print(f"{log_name} Retrying in {wait_time} seconds...")
+                    time.sleep(wait_time)
+                else:
+                    print(f"{log_name} All connection attempts failed")
+                    raise
+
     try:
         check_python_version(infile)
 
@@ -68,8 +90,9 @@ def process(df_id, batch_id):  # type: ignore[no-untyped-def]
 
         # To attach to the existing SparkSession, we're setting the session_id in the URL.
         connect_url = connect_url + ";session_id=" + session_id
-        spark_connect_session = SparkSession.builder.remote(connect_url).getOrCreate()
-        assert spark_connect_session.session_id == session_id
+        
+        # Use retry logic for better reliability in CI environments
+        spark_connect_session = create_spark_session_with_retry(connect_url, session_id)
         spark = spark_connect_session
 
         func = worker.read_command(pickle_ser, infile)