"Stateless Linux" project [LWN.net]

"Stateless Linux" project

Posted Sep 14, 2004 9:53 UTC (Tue) by job (subscriber, #670)
Parent article: "Stateless Linux" project

The PDF is 9 pages of technical marketspeak. They pretend mounting r/o NFS root is something new which sort of bothers me. Have they even read the LSB? After all, that mounting is why we separate architecture dependent /usr/bin from independent /usr/share.

But the project seems legit, and very ambitious. They plan to make it possible to 1) use and configure USB PnP hardware on a machine with r/o root and NO root privs, 2) set date+time etc. without root privs and 3) suggest a good network file system that caches (like disconnected AFS or CODA).

I often mount root from several clients, and upgrade the system from the server, as long as hardware is identical. This approach would make that more useful. It is good. (Only the paper, written for windows people and needs deciphering to read, is bad.)

(Log in to post comments)

"Stateless Linux" project

Posted Sep 15, 2004 10:23 UTC (Wed) by hppnq (subscriber, #14462) [Link]

Maybe you are joking, in which case you can ignore this comment.

The comment is pointless in itself, and you are not doing Havoc justice. I am not even going to give you pointers to the numerous contributions he has made to the Open Source community. Think twice before you judge someone like that in public, if you don't want to look like a complete idiot.

Have you actually read the paper?! Where exactly is the "marketspeak"?

We need more people who are really good at the technical stuff and who are also able to tell others about it.

(And why is it that people have decided to aim their mindless rants at RedHat all of a sudden? Spare your breath for the real thing, mate, and go bother someone else in the mean time, if you please.)

"Stateless Linux" project

Posted Sep 15, 2004 11:37 UTC (Wed) by hppnq (subscriber, #14462) [Link]

(And why is it that people have decided to aim their mindless rants at RedHat all of a sudden? Spare your breath for the real thing, mate, and go bother someone else in the mean time, if you please.)

Sorry for replying to myself, but maybe I should clear this up a bit. First off, smoogen, the above remark was not aimed directly at you.

What annoys me is that the number of comments on LWN that fall in the category of ranting attacks on RedHat has risen to an alarming high, or at least so it appears to me. Is there something I missed?

After the whole discussion about Fedora's (lack of) interaction with the community we now get a proposal for the community to play with and a nice document explaining this proposal; I would think this is A Good Thing (TM).

Just drop your tinfoil hats for a moment, read the stuff, think about it, relax your trigger-happy mouse finger, say your favourite mantra, and then start thinking about how to word a comment if you have something to add to the discussion. I promise you I will do the same. ;-)

(Apologies to anyone who is fed up with having to wade through comments that are not actually adding anything to the discussion. I am with you. I like the excellent articles here at LWN, and the comments; there are some real gems every now and then. The last thing I want to do is add noise. ;-)

"Stateless Linux" project

Posted Sep 23, 2004 15:42 UTC (Thu) by job (subscriber, #670) [Link]

So your comment was directed at me then. The "marketspeak" is what the
paper is full of. "Marketspeak" because it is indeed correct but not
written in the terse technical sense we are used to, especially from Red
Hat.

There are no hats involved here, neither tinfoil nor red ones. That I, or
anyone else, should choose not to comment on it just because it is from
Havoc or Red Hat itself is something netiher of us wants, if you think
about it.

Obviously I read the paper since I summarized it above. Did you read my
comment?

"Stateless Linux" project

Posted Sep 14, 2004 15:00 UTC (Tue) by hp (subscriber, #5220) [Link]

I don't see how the paper pretends NFS mounting is something new, since it points out that LTSP already offers that on the first page. And the conclusion section has a long list of existing projects that already offer aspects of this, including mentioning LTSP again, and MIT Athena which is also based on read-only root.

The point is not that it's some kind of 100% revolutionary discovery, but that a long list of specific stuff needs doing in Fedora specifically to enable this functionality out of the box, in a nice clean framework, with UI that supports it (e.g. system-config-* tools that edit config files in /etc are not the right approach for desktops if we assume read-only /etc). There's a whole long list of implementation work to do.

The name "stateless Linux" is invented to highlight that diskless, "cached client", and boot-from-CD systems can be handled in the same framework; and to emphasize that the problem is larger than just getting tftp/pxe to work, you also have to change the OS extending up to the desktop user interface.