GIT_SSH script file handle not yet closed when clone command issued.

[croth1]

Hi, this code fragment somehow can have unexpected behaviour on (some?) linux machines. For me subprocess.check_call fails, because on the OS side, the file handle to the temporary file has not been closed yet (despite the explicit .close() statement) - debugging the check_call leads to a Text file busy error message).

sagemaker-python-sdk/src/sagemaker/git_utils.py

Lines 281 to 287 in ac353b2

	with tempfile.NamedTemporaryFile() as sshnoprompt:
	write_pipe = open(sshnoprompt.name, "w")
	write_pipe.write("ssh -oBatchMode=yes $@")
	write_pipe.close()
	os.chmod(sshnoprompt.name, 0o511)
	my_env["GIT_SSH"] = sshnoprompt.name
	subprocess.check_call(["git", "clone", repo_url, dest_dir], env=my_env)

Interestingly, the code works fine on my mac, so the behaviour seems somewhat OS dependent.

Instead, changing the NamedTemporaryFile to a TemporaryDirectory and creating the helper script within the context manager of the TemporaryDirectory should work, I think.

[croth1]

This fixes the problem for me:

diff --git a/src/sagemaker/git_utils.py b/src/sagemaker/git_utils.py
index 80bd62d5..62752bb2 100644
--- a/src/sagemaker/git_utils.py
+++ b/src/sagemaker/git_utils.py
@@ -14,6 +14,7 @@
 from __future__ import absolute_import
 
 import os
+from pathlib import Path
 import subprocess
 import tempfile
 import warnings
@@ -278,12 +279,13 @@ def _run_clone_command(repo_url, dest_dir):
         my_env["GIT_TERMINAL_PROMPT"] = "0"
         subprocess.check_call(["git", "clone", repo_url, dest_dir], env=my_env)
     elif repo_url.startswith("git@"):
-        with tempfile.NamedTemporaryFile() as sshnoprompt:
-            write_pipe = open(sshnoprompt.name, "w")
-            write_pipe.write("ssh -oBatchMode=yes $@")
-            write_pipe.close()
-            os.chmod(sshnoprompt.name, 0o511)
-            my_env["GIT_SSH"] = sshnoprompt.name
+        with tempfile.TemporaryDirectory() as tmp_dir:
+            custom_ssh_executable = Path(tmp_dir) / "ssh_batch"
+            with open(custom_ssh_executable, "w") as pipe:
+                print("#!/bin/sh", file=pipe)
+                print("ssh -oBatchMode=yes $@", file=pipe)
+            os.chmod(custom_ssh_executable, 0o511)
+            my_env["GIT_SSH"] = str(custom_ssh_executable)
             subprocess.check_call(["git", "clone", repo_url, dest_dir], env=my_env)