On December 15, 2011 12:39:59 PM Josselin Mouette wrote: > Le vendredi 16 décembre 2011 à 03:35 +0800, Thomas Goirand a écrit : > > Oh, and when I'm at it, how do you implement /usr as read only, > > (over nfs for example)? This is a quite common setup in large > > organization / universities. > > No, it is not. With a packaging system it is not suitable to have a > NFS-shared /usr.
hmm... With dpkg's --path-exclude option it is now possible to fairly efficiently install /usr/* onto only one box on the LAN. AFAICT, the only major thing missing is a way to selectively and simultaneously push upgrades onto other boxes--which I suspect would be do-able with a dpkg-daemon and some coordination. Even though disk space is cheap nowadays, it seems like a waste of 10's of GB to duplicate essentially static files on multiple systems. There is also the extra admin time it takes to upgrade multiple systems vs. upgrading one and having the other boxes automatically follow along. The use cases I'm thinking of are those where the same software is wanted on each box but configs may vary between them, such as households with multiple computers or offices with more than a couple secretaries and sales staff. I'm also thinking this is the kind of feature which there is not currently much demand for simply because it is traditionally the domain of mainframe+terminal systems (i.e., it wasn't considered possible with desktop boxes). > OTOH it is a common setup to share / over NFS. Unfortunately, when / is imported over NFS the box is effectively a door stop when networking fails, and if it doesn't have an optical drive you can only troubleshoot from one end of the connection. - Bruce (who files bug reports when his rare system fails to boot because some NFS stuff lives on the imported /usr partition) -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/201112221411.22879.bms...@shaw.ca
