I want to inject a switch to a private mount namespace into the remote login process, so the remote user's desktop and everything under it is running under that namespace.

I've only just read about private mount namespace (while searching for a solution to another problem). I've never used the feature, so any tutorial or cookbook instructions for any use of private mount namespace would help.

Having discovered this feature exists, I'll probably want to use it for a few things, but at the moment my objective to make most of the hidden files and subdirectories in a user's home directory differ depending on whether the user is logged in on the local display or logged in remotely.

My basic plan (please tell me if there is a better way or if there is a serious flaw) is to replace each of the .* names in /home/fred with a symlink to the same name in /local/fred those would be the files used if fred logs into the physical display.

Now if I can get remote log ins to use a different private mount namespace, I would mount a directory on one of our fileservers in place of /local in the private mount namespace of remote users. That would contain a fred subdirectory with a different version of those hidden files, a version used whenever fred logs in remotely to any Centos 6.5 in our LAN.

Any constructive suggestions, alternatives, etc. would be appreciated. I'm way out of my depth here.

Centos 6.6
Kernel 2.6.32

My purpose for all this is explained better in my related thread:
http://www.linuxquestions.org/questi...06#post5088506