-----BEGIN PGP SIGNED MESSAGE----- [EMAIL PROTECTED] (joost witteveen) writes: | | > As near as I can tell from examining various FAQs, there is no support | > for maintaining clients with a read-only, NFS-mounted "/usr" | > partition. The only current solution seems to be: (i) install a | > complete system on the server and a base system on the client; | > (ii) remove the client "/usr" partition and mount the server "/usr" in | > its place; (iii) configure the client's packages by hand. | | Well, as far as I (and about 5 happy clients (who mount /usr from | my servre) can see, you're somewhat wrong.
I don't mean to be ungrateful, but your message doesn't shed much light on the issue. Perhaps you could supply a bit more detail? How do you install/configure packages on the client side? How do you synchronize package installation and upgrades between the server and the client? How does your solution differ from my proposed (and in my opinion, pretty bad) solution---that is, how do you avoid configuring the client's packages by hand? | The only trouble is /etc: it has to be mounted rw (/etc/mounts, and | /etc/ld.so.cach), and has to have some files different from | the server (like /etc/init.d/network, setting up different IP numbers). | | But on the /usr side, its all _very_ easy! Well, yes, but I already knew that. I want to know what you *do* for the client "/etc" partitions? Do you run a nightly script to update the client's "/etc"? How does your script determine what should and shouldn't be overwritten in "/etc"? Kevin <[EMAIL PROTECTED]> -----BEGIN PGP SIGNATURE----- Version: 2.6.2 Comment: Processed by Mailcrypt 3.4, an Emacs/PGP interface iQBVAwUBMhh+Y4mVIQW1OgXhAQFDjQH+Ku6bRO0OzI31PG9wWJnJR/kjGAVl8asf KwErp70Yodj9vxNw0eWljREEO5J3cDHYiy4vzOyVdj8CbiCQvGUNOA== =+P2R -----END PGP SIGNATURE-----
