k8s: reana user available system wide

[Sinclert]

Problem

When running an analysis that needs to obtain the user uname using the getpass Python library (pwd under the hood), the reana user that gets plugged by the workflow controller just before executing the desired workflow step, is not available neither within the /etc/passwd nor the /etc/shadow files.

Code snipped plugging reana user:

reana-workflow-controller/reana_workflow_controller/workflow_run_manager.py

Lines 560 to 579 in 4c59ec8

	def _create_job_controller_startup_cmd(self, user=None):
	"""Create job controller startup cmd."""
	base_cmd = "flask run -h 0.0.0.0;"
	if user:
	add_group_cmd = "groupadd -f -g {} {};".format(
	WORKFLOW_RUNTIME_USER_GID, WORKFLOW_RUNTIME_USER_GID
	)
	add_user_cmd = "useradd -u {} -g {} -M {};".format(
	WORKFLOW_RUNTIME_USER_UID, WORKFLOW_RUNTIME_USER_GID, user
	)
	chown_workspace_cmd = "chown -R {}:{} {};".format(
	WORKFLOW_RUNTIME_USER_UID,
	WORKFLOW_RUNTIME_USER_GID,
	SHARED_VOLUME_PATH + "/" + self.workflow.workspace_path,
	)
	run_app_cmd = 'su {} /bin/bash -c "{}"'.format(user, base_cmd)
	full_cmd = add_group_cmd + add_user_cmd + chown_workspace_cmd + run_app_cmd
	return [full_cmd]
	else:
	return base_cmd.split()

---

How I ran into the issue

I was designing a workflow that uses MLFlow to send information to a server to track specific information across multiple runs. When running the workflow, I encounter the following error:

  ...
  File "/usr/local/lib/python3.8/dist-packages/mlflow/tracking/context/default_context.py", line 19, in _get_user
    return getpass.getuser()
  File "/usr/lib/python3.8/getpass.py", line 169, in getuser
    return pwd.getpwuid(os.getuid())[0]
KeyError: 'getpwuid(): uid not found: 1000'

---

Possible solution

When looking for similar problems online, I found a couple of GitHub discussions about:

• How useradd works on Alpine / Ubuntu.
• Why a user UID could be missing (the merged PR listed below is particularly useful).

I think the issue could be solved by using the -r/--sytem flag on the useradd command, as the Linux manual describes.

---

I am not that familiar with the reana-workflow-controller, but I think it is a good guess.
Any feedback will be welcome 😄