With no native file sharing feature in the Virtual Machine Manager, guest systems must rely on standard network file sharing to access files on the host storage device. For systems that would perform automated tasks without user login, on files in the shares of the network device, this demand creates a dilemma, because file sharing must be permissive enough to allow mounting by the guest system without a user supplying a password, but also must be restrictive enough to enforce adequate security.
For provisioning a Linux client for automatic mounting of shared volumes, it would seem that common solutions fall into one of three categories:
Option (2) might be slightly more secure, but storing credentials, especially a string created by the user, and persisted in plain text, is cumbersome and vulnerable. (I believe that in this case the password would be accessible also to any regular user of the system, if not through reading the the fstab file, then by listing the table of active mounts.)
Option (3) solves most of the above problems, but incurs administrative overhead exceeding that feasible for most small deployments, at least in the case specifically of Kerberos, which DSM supports.
Are any better options overlooked, perhaps more secure than (2) but less difficult to manage than (3) for a small but secure deployment of an automated client system?
For provisioning a Linux client for automatic mounting of shared volumes, it would seem that common solutions fall into one of three categories:
- Mounting an NFS share that is accessible without password, perhaps with the IP address of the client system serving as the only means of authentication.
- As above, but using a password stored on the client.
- Using a key-management authentication and authorization system, such as Kerberos.
Option (2) might be slightly more secure, but storing credentials, especially a string created by the user, and persisted in plain text, is cumbersome and vulnerable. (I believe that in this case the password would be accessible also to any regular user of the system, if not through reading the the fstab file, then by listing the table of active mounts.)
Option (3) solves most of the above problems, but incurs administrative overhead exceeding that feasible for most small deployments, at least in the case specifically of Kerberos, which DSM supports.
Are any better options overlooked, perhaps more secure than (2) but less difficult to manage than (3) for a small but secure deployment of an automated client system?