Fedora Project Considering "Stateless Linux" 234
Havoc Pennington writes "Red Hat developers have been working on a generic framework covering all cases of sharing a single operating system install between multiple physical or virtual computers. This covers mounting the root filesystem diskless, keeping a read-only copy of it cached on a local disk, or storing it on a live CD, among other cases. Because OS configuration state is shared rather than local, the project is called 'stateless Linux.'
The post to fedora-devel-list is here, and a PDF overview is here."
Looks neat but... (Score:4, Interesting)
Re:Looks neat but... (Score:5, Interesting)
You can still have one user work and experiment on a kernel module and crash his system while another continue with her wordprocessing.
Re:Looks neat but... (Score:5, Informative)
Re:Looks neat but... (Score:5, Interesting)
Re: (Score:2)
Re:Looks neat but... (Score:5, Informative)
If this thin client cluster idea appeals to you, please see ltsp-mosix [lpmo.edu].
Re:Looks neat but... (Score:4, Informative)
Once such a system is set up properly, it could be self maintaining with a significant reduction in hardware and energy and maintenance costs.
Re:Looks neat but... (Score:2, Funny)
Compute farm / beowulf (Score:2)
That description matches an compute farm in the next room [0]. It also handles the case of 'diskless' install with a local disk, used for application specific working space [1].
Hell, in the next building there is a beowulf of 32 nodes that hasn't bee updated because the updating of 32 nodes wasn't automated, and time crunch [2]. If it's all from a single image, that's trivial to up
Re:Looks neat but... (Score:2)
Setting up a linux 'terminal server' (using XDMCP to provide X logins to thin clients) is exceedingly easy [google.com] to set up, and your thin clients can be running pretty much any UNIX flavour that supports XDMCP. I personally like this set up because the client computers can be as dumb as possible (and bloody cheap), and you can invest server resources in your central server - make it real beefy, dual processors, gigabytes of RAM :) The t
The logical conclusion (Score:4, Interesting)
A file/directory is either
- Static (not changed except by action of the system administrator), or
- Variable (subject to change at any time
and either- Shareable (multiple machines can have a common copy), or
- Unshareable (each machine needs a separate copy).
In an effort that is conceptually equivalent to the separation of the kernel tree into architecture-dependent and -independent subtrees for the Alpha port, which made subsequent architectures far easier, a lot of people have devoted their efforts to determining just how little of what goes into the file hierarchy really has to be unique to the machine.The 'aha moment' comes when you think of groups of workstations with identical hardware, which are candidates for having a common image from which they can be built, and realize that you can build a relational database that correlates MAC addresses (possibly to some other locally-unique but shorter machine number) to the HW configuration. Now, conceptually all of those cookie-cutter-identical machines are a single entity for the purposes of configuration. A lot of what FHS considers 'unsharable' is now quite 'sharable' within such a HW config group.
As workstations age, the IT department brings in a couple samples of the next HW configuration, loads drivers, tests against the app suite, and when they're ready for primetime, the vendor delivers them, the MAC addresses are added to the database, the workstations boot up, find Mommy (bootp server), and Just Work. The user can log out of an old computer and into a new one, and find all his 'stuff' right where he left it. It's the only sane way to compute in an institutional environment.
NFS Mount? (Score:2, Interesting)
Re:NFS Mount? (Score:5, Interesting)
If you'd bother to read the white paper or howto (sure, I'm new here) you'd have read that this is more than NFS mounted roots.
It's a framework for managing the servers, cached operation, integrated authentication etc. You can use this framework to manage roaming devices like laptops, allowing automatic install images, etc. etc.
An NFS solution requires network connectivity the whole time, this doesn't.
LTSP (Score:3, Interesting)
Thin clients WOULD be a blessing, I imagine. Single configuration, one update, all the "personal files" in a server somewhere -- makes for easy updating and backing up. Also keeps hardware requirements down...which [buzzword warning] "helps lower TCO and increase ROI"
Re:LTSP (Score:5, Funny)
Welcome to the world of 'dumb terminals' again. Thanks for playing this long!
Re:LTSP (Score:2)
Quite right. The move in business IT from centrally managed mainframes to networked PCs was a huge step backwards in terms of cost and availability of line-of-business applications.
The average user, once they learned how to use a particular application, never had to worry about IT, because it "just worked". Contrast this to downloading patches every other day, running weekly virus scans, keeping your PC current in the cor
Re:LTSP (Score:5, Informative)
Seems like a good idea to me.
Re:LTSP (Score:4, Interesting)
This kind of disconnected caching would be excellent. In some ways it's a kind of uber-sync.
What fedora is experimenting will work great on thin and thick clients. I think this is an exciting development, and even for maintaining just a few machines around the house would be nice to have that kind of capability.
Also, I would say that yes, thin clients are coming back into fashion. But thick clients are here to stay also.
mainframe (Score:4, Interesting)
Re:mainframe (Score:2)
Until the central server crashes and nobody can do anything.
Re:mainframe (Score:4, Insightful)
Re:mainframe (Score:2)
Re:mainframe (Score:3, Interesting)
Part of the problem is that while I don't trust users to keep their machines running properly, I barely trust a lot of server admins to do any better. I've seen the way a lot of servers are put together, and how often they need some really inane maintenance. It's scary. The penalty for a bad user is usually limited to affecting one or two people
Re:mainframe (Score:5, Informative)
And since we cannot do without networking anyway, and since storage devices are easy to make high available, this would seem like a blessing to me.
Re:mainframe (Score:2)
On behalf of non-geeks, let me be the first to... (Score:3, Funny)
I mean, I know the words. It's mostly English, and that's my first language, and I'm pretty handy with computers, but that was the most incomprehensible load of babble I've heard since the last time I watched TNG.
Can someone explain what this means, in plain English, to a regular user (i.e. non-hacker geek types)?
Re:On behalf of non-geeks, let me be the first to. (Score:5, Informative)
thats it in a nutshell....
Re:On behalf of non-geeks, let me be the first to. (Score:3, Funny)
Very simple, it is stateless so it remembers nothing from command to command. Here's what it would look like to use it:
I for one plan to skip this distro.Wow! (Score:4, Insightful)
And please all the "NFS root is enough" posts - read the article!
Re:Wow! (Score:2, Insightful)
Re:Wow! (Score:2)
That's the problem (Score:2, Insightful)
Re:That's the problem (Score:2)
What about the ones who haven't learned Linux a first time yet? Might they find this useful?
You're right It's ambitious. I don't think that is a bad thing though.
Re:That's the problem (Score:5, Insightful)
Well, most of us don't /really/ want to relearn *anything*. Sometimes, however, when you hear a new idea relating to an area you work in, the penny drops, and you are left thinking "wow, what a great idea".
For instance, I work in a scientific research environment (high energy physics) where most of our software is Free (capital F), we work in different places at different times (planning, lab, analysis), we have a great deal of customized and hand written software and the ideal development environment so far has been NFS mounted home directories (running RedHat and now Fedora). In theory every machine I log into is running the same OS, with /usr/local NFS mounted from an [application|file] server, I login though NDIS and my home directory is also NFS mounted.
This works fine in theory - except without a serious admin budget, different OS versions spring up... I have access to machines running RH9, FC1, FC2... and that's an improvement, whilst RedHat were still supporting RHL, we had 7.3, 8.0 and 9.0, with wildly different GCC versions. What happens? I end up using specific machines with a similar enough environment that all my simulations will at least compile without tweaking, and all my scripts etc work the same way. Homogenous environments, no matter how ideal, are not a possibility without a manpower commitment that many SMBs and other small operations can't afford.
This stateless project LEAPS out at me as an ideal way for small operations (like up to 100 seats) to be managed by a single (even part time) admin.
Not to mention the attempt to tackle laptops - which is the reality of the workplace. Many people have laptops. A lot of them (and their CTOs) would love to be running the same environment as the workplace LAN. At my lab most people have a laptop due to the amount of travelling we do - I'd guess that 90% of them are running XP, since even if they did run linux, they'd have to administer it themselves, wouldn't have clearance to access the NFS shares for $HOME and /usr/local.
Although the laptop aspect still has a troubling achilles heel: most of us (well, my colleagues at least) have laptops in order to present our work to others. Even ignoring the ubiquitousness of PowerPoint, who amongst us would want to be on the road with a "cached client" laptop with NO write-access to anything but $HOME. Sure, the system worked at the office, and you fixed all the bugs that cropped up when you connected from home on you DSL, but what about a strange environment. You need to connect over someone elses WiFi to get the latest figures (sure, TFA talked about user-configured WiFi, but still, what if they have different security like WEAP that needs a new package and root access), or if you NEED to plug in a USB key to give a collaborator or customer your files. What then?
Regardless, this to me is a prospective Killer App for linux, and is definitely tackling a bunch of issues that may niggle an admin for several years before they could even define what the problem is. Automatic updates across _all_ your workstations. Backups that require 10 minutes work after a crash - and I can attest that a recent HD crash to our "distributed" system took a few hours to get the machine back together, but several days before all the little minor tweaks we needed had been applied (things like monitor resolution, 'sudo' configuration, extra packages, sound drivers.
For the first time, I stand up and say, THANK YOU REDHAT and THANKS FEDORA. This project tells me that you are thinking about your installed customer base and offering _really_ innovative ideas to the community. Anyone want to moan about how Linux is always playing catchup to MS and Apple and how F/OSS is doomed to lag behind forever?
I want the opposite! (Score:5, Insightful)
It's really disconcerting for me that practically all the distros want you to have root access even to install a simple MP3 player from their package files; and extremely distrubing that they do it by popping up KDE or Gnome windows asking for root paswords.
Isn't this what we blame microsoft for?
Disk space is cheap enough, we don't need more sharing of config stuff - we need more separation so users can use the benefits of package managers without having to get in the way of other users.
Re:I want the opposite! (Score:3, Interesting)
Re:I want the opposite! (Score:2)
Re:I want the opposite! (Score:2)
Re:I want the opposite! (Score:2)
Re:I want the opposite! (Score:2)
Re:I want the opposite! (Score:5, Informative)
Not quite: we blame them for having to *run* a lot of programs as root to get full functionality. In most *nixes, OTOH, you only need root passwords to *install* programs, while the programs themselves run just fine for regular users.
I dont see anything wrong with having to ask for root passwords for critical changes to any system: its a good practice, and one of the better implementations of it is seen in OS X, which actually has 'Lock/Unlock' icons for settings that need root access.
Re:I want the opposite! (Score:5, Informative)
Take a look at zero install [sourceforge.net]. You can install 0install on many distros (as root) then install apps as a user exactly like you want.
Or buy a mac!
Re:I want the opposite! (Score:2, Insightful)
Here's a few points. First of all, you can configure KDE or Gnome not to ask. Second of all, most users are not admins. Allow me to expand on that. Most people who use computers have no idea of what is harmful and what is not harmful and will install anything. Theoretically the admin should install the basic apps (office, music, and internet) so that users won't go a
Re:I want the opposite! (Score:2)
The only problem with installing a package under $HOME is that software generally expects things to be in certain directories, unless you compile them, and then you can build them in and install them to your hom
Re:I want the opposite! (Score:4, Insightful)
MOD THIS UP, GUYS! :-) (Score:2)
Paul B.
Re:I want the opposite! (Score:5, Insightful)
Were the internet a safe place, I'd almost agree with you. Almost.
Isn't this what we blame microsoft for?No. I've never blamed MS for this, who by default, logs in users as administrators. Which is a terrible idea, security wise, and they've been pulled over the coals several times for it. Rightly so.
Disk space is cheap enough, we don't need more sharing of config stuff - we need more separation so users can use the benefits of package managers without having to get in the way of other users.
No, what we need is users to do their job and stop trying to get around the restrictions the admins put in place, which is exactly what your idea would be used for.
In fact, in all my production systems, home is ALWAYS mounted as noexec. You want a program on the server, fine, you let me know which one and why, and I'll think about it.
Re:I want the opposite! (Score:4, Insightful)
Requiring the root password for certain tasks does not increase security, IMHO. Most users (a) don't want to be constantly typing in passwords and (b) would type it in whenever it was asked for without thinking too hard about it.
If anything you don't want the typical personal-PC one-user setup to ask for the root password very often because the more often you ask when it's not really needed, the greater "password fatigue" gets and the less likely people are to think critically when they get asked.
Really, if you spend a lot of time thinking about it as I have, you come to the realisation that malware which relies on social engineering doesn't have any useful technical solutions. You can get some way there with things like distributed whitelists but pretty quickly you end up in the realm of civil liberties (who really owns that machine you paid for?).
In short: making tasks hard doesn't increase security, it just annoys the user. If the user has decided they want to do something, they'll do it. So good security in the face of a dangerous net is about advising the user well whilst not getting in their way.
Now, I know you're coming from the viewpoint of a server admin which is fine. Most people aren't server admins. It's wrong to try and apply the tools used to admin servers to home machines.
That's one reason why autopackage can install to home directories. [autopackage.org] (see the third screenshot), though it's not really something that's encouraged (and it can be disabled by administrators). Another is because it's really useful if you want to use a newer/older version of the software installed on a multi-user machine without interfering with other users. Another is because some shell accounts do let you run programs and it's nice to be able to use binaries.
In fact, in all my production systems, home is ALWAYS mounted as noexec. You want a program on the server, fine, you let me know which one and why, and I'll think about it.
That doesn't help very much, you can still run programs on a no-exec mount if you really want to.
Re:I want the opposite! (Score:4, Insightful)
$ chmod -x ~/yes
$ ~/yes
bash: ~/yes: Permission denied
$
y
y
You might wonder how this works?
Re:I want the opposite! (Score:2)
I think that pretty much sums up why client/server computing is dead and everyone runs a local copy of Windows as admin.
Re: I want the opposite! (Score:2)
The first user you create is an admin, but runs (kind of) as an 'ordinary user', but with the power to 'sudo'. Think of them as being in the 'wheel' group.
If I want to drag an application from an install disk image (or compile one) I can. If I want to make it usable by all users, I need to enter my password when I drag it into the 'global' applications folder.
OK, some apps require the admin password to install (and many shouldn't, but still do it)
Like Clusters (Score:5, Interesting)
Again... (Score:5, Insightful)
NFS read-only & shared root is enough
+
LTSP
+
Thin clients
=> please read the article
Re:Again... (Score:2)
!(NFS read-only & shared root is enough+LTSP+Thin clients) OR please read the article
I hate college
A few thoughts (Score:3, Insightful)
First, what's so special about this? If you set up a network filing system for your root FS and use LinuxBIOS as your bootable image, you can have a single, central Linux install that is shared with as many computers as you like.
What would be far MORE interesting would be to have a central server with multiple images for different hardware. Then you could boot your nice, shiny IBM mainframe from the same "install" as your desktop PC or the webmaster's Apple Mac.
Another possibility is a massively parallel installer. Basically have one machine on which you are actively installing, but have that machine replicate the write-to-disk operations across the entire network to all the other PCs.
A third option would be to have a distro which set up the entire network as a cluster, but with the config files just on one machine. That way, you don't burden any one machine with having to serve the really heavy-duty stuff, such as applications.
Re:A few thoughts (Score:2)
What's the fastest way... (Score:2, Insightful)
Wow, this looks handy (Score:2)
Back to mainframes? (Score:5, Insightful)
Re:Back to mainframes? (Score:2)
Re:Back to mainframes? (Score:2)
RTFA, dammit! (Score:5, Informative)
From the article:
Re:RTFA, dammit! (Score:2)
Sounds like the Coda or Intermezzo filesystems to me.
I was just thinking AFS.
Unison is another option.
With a combination of unison, AFS/Coda/Whatever and/or a local partition, I could see doing this under linux.
Not needing root and thin client hybrid... (Score:5, Informative)
1) they don't want users to need root for hardware (but do want users to need the admin to install certain software). This info is in the PDF. They already see that needing root for hardware install or configuration needs to be worked around.
2) the design is a hybrid or amalgamation of thin and fat client, trying to cherry pick the best of both:
applications run on local systems
software and data cached on local disk
central management and configuration of nodes
they call it a cached client technology
3) they have a plan for laptops. Stateless... instantiation, sync... things that sound vague, but they seem to have a plan because this stuff is considered in the howto. There are some notes in the how-to covering the different types of clients:
" diskless clients, which boot directly from a snapshot stored on the server
caching clients, which boot from a copy of a snapshot, cached locally on a hard drive.
Live CD clients, which boot from a copy of a snapshot burned onto a CD
thick clients, which don't use snapshots and must be maintained by another means.
"
The idea has some very cool potential for a business or network situation. I can't imagine this is ready for production, but it could be soon.
-A
heading off the misinterpretations (Score:4, Interesting)
First of all, I'm not associated with the project.
However, I've read what they're talking about, and here is where many people are misinterpreting:
This is not a 'thin' client in the traditional sense. The client in this case does the computations.. i.e. it actually runs the app.
In other words, the computer is not merely a display, and as such shouldn't suffer from the traditional mainframe/client shortcomings.. (you have all the CPU power you normally have)
When you think about this, think KNOPPIX and other live-cds, that is the nearest (and quite near, imho) to what they're discussing.
So... why is this different from a normal install?
A normal install has a read-write root, whereas here they're shooting for a read-only root, even if it is still on the local harddrive.
Answer to the SCO issue (Score:3, Funny)
Sure, ping times will be a bitch, but...
This addresses a real problem... (Score:2)
My girlfriend has a laptop from work, a large company that enforces a "users don't get admin access to their machines" policy. Fine and dandy, until she brings the laptop over to my house and wants to print something on my printer. Whoops! No device driver for that particular kind of printer in the standard corporate install, and even though I
Innovation! (Score:2, Troll)
But the one thing [anti-linux] people keep saying is that Linux is all about being a copy-cat and nothing about innovation, new development new technologies or new ideas.
Recently, along with this and so
Re:Innovation! (Score:2)
If you think that only "recently" linux and OSS have begun to innovate then you've been living with blinders on.
Good framework for future development (Score:3, Insightful)
Most importantly, this
1. avoids the absurdity of moving all processing, and indeed disk to a central server
2. focusses attention on development and maintenance of prototype installations for different types of machines
Some of the implementation techniques don't seem pleasant--but they're doing things in a way that appears forward-looking.
I look forward to seeing more of this.
Sounds like X terminals to me (Score:2)
Actually its the only way to fly in an enterprise environment.. Get the PC back out of the users hands. Should never have given them to the users in teh first place.. 3270's for all!
sounds a lot like VAXcluster (Score:2)
Everything old is new again!
The Exodus of Technologies to the OSS Realm (Score:2)
But you don't have to pay DEC an arm and a leg for hardware and software licenses.
The migration of all these ideas into the realm of industry standard, commoditized hardware is a huge deal. The age of one company being able to own a whole market vertically, from the silicon to the user interface, is gone. We left those monopolies behind, and good riddance. But we also left some good ideas behind with those no
Got to love stateless installs (Score:5, Interesting)
It had / mounted read-only.
You could power down the thing whenever the hell you liked and never see fsck run.
Local cached copy of filesystem (Score:4, Interesting)
I think there should be a more general concept of overlayed filesystems, where a FS could be mounted on top of another FS "with transparency", so that you can see all the files in the entire "stack". A standard "ls" would show 1 instance of each file, with the "highest level" FS taking precedence. A modified program might be able to see all the versions of a particular file and be able to copy one to another (if permissions allow).
If each FS could be mounted RO or RW, then you could have a local copy of everything on a CD or DVD, but make it appear writable by mounting another FS on top (either a local HD, USB pen drive, NFS mountpoint, etc). Recovering back to the original install would be just wiping out the modified files, so the underlying files are now visible.
This would be good for:
- fully functional Linux systems based of a CD or DVD
- FS snapshots for backup or testing
- intrusion detection (diff across file versions)
- version control of the entire OS image
Now, if only I were smart enough to actually write the code.
Re:Local cached copy of filesystem (Score:2)
I think there should be a more general concept of overlayed filesystems, where a FS could be mounted on top of another FS "with transparency", so that you can see all the files in the entire "stack". A standard "ls" would show 1 instance of each file, with the "highest level" FS taking precedence. A modified program might be able to see all the versions of a particular file and be able to copy one to another (if permissions allow).
Well, I'm not sure about your operating system, but this layout is one I c
Re:Local cached copy of filesystem (Score:2)
Peer OS? (Score:2, Interesting)
Separate the state from the behavior with respective hardware, sounds interesting. Definitely they will need to break all the encapsulation layers built in todays modern OS and identify the patterns that represent common behavior and common state.
In the article, it makes me wonder, is it better to centralize state or behavior? For instance, centralizing state would be more efficient, but if state
An amazinq new innovation! (Score:2, Informative)
Interesting project (Score:4, Interesting)
'Thin client' was the first attempt to dethrone MS in this way, but this approach appears much more sophisticated, and consequently much more likely to succeed. Without seeing how the whole thing plays out I really have no idea whether the approach is successful or not. But it's a really nifty shot across the MS bows.
Whether this goes anywhere or not ends up being decided by (as with most IT projects) whether the services provided by IT to the end users are adequate (in which case IT gets their way) or so obnoxiously limited that the end user cabal ends up storming the IT department with burning torches.
stateless? (Score:3, Insightful)
Marketing = liars, even at Red Hat (Score:2)
http://bugzilla.redhat.com/bugzilla/show_bug.cg i ?i d=119185
If you follow the link and you can't believe what's recorded there, it's still correct: if
Re:Marketing = liars, even at Red Hat (Score:3, Informative)
The bug was closed as WONTFIX because the reporter was an obnoxious prick. Referring to the developer as a Moron on repeated occasions. The fact is that if you want people to help you, yelling abuse is not a particularly good strategy.
Re:Marketing = liars, even at Red Hat (Score:2)
The bugzilla entry is not a request for technical expertise, it's a bug report. It doesn't matter if the submitter is a socially challenged idiot.
The DRBL (Diskless Remote Boot in Linux) project (Score:2, Interesting)
http://drbl.nchc.org.tw (Traditional Chinese)
and
http://drbl.sf.net (English).
Maybe someone can have a look at that, some part of DRBL are similar to this Stateless Linux project.
DRBL runs well on RedHat, Fedora, Mandrake and Debian.
In Taiwan, more than 100 sites already downloaded and run DRBL, some of them are schools (Primary/High school/University), some of them are NPO and buisness companies.
check this:
http://drbl.nchc.or
Looks like OsX (Score:3, Interesting)
I really don't see the point (Score:4, Interesting)
Re:Until they fix the license (Score:3, Interesting)
Re:Until they fix the license (Score:2, Offtopic)
Ahhh, but can anyone really be born gay?
I didn't know what Astroturf was.. (Score:2, Informative)
---
In American politics, the term astroturfing is used perjoratively to describe formal public relations projects which deliberately give the impression of spontaneous and populist reactions.
The term is a play on "grassroots" efforts, which are truly spontaneous undertakings. AstroTurf refers to the bright green artificial grass used in some indoor sports stadiums.
A "grassroots" action or campaign is one that is started spontaneously and is largely
Information please! (Score:2)
Personally, I use Fedora myself and enjoy it. I would hate to discover that RHAT is employing such an underhanded tactic.
Re:Information please! (Score:2)
Re:What's wrong with flexibility? (Score:3, Funny)
And you'd have been correct.
Now this totally neglects the less-than-common knowledge that they were actually created in America in the 1800s by immigrants to mining communities as a means of differentiating their restaurants from more common fare
Crap. Chopsticks have been in use in China and Japan for around 5000 years. This page [calacademy.org] includes a brief history, and you can get m
Re:What's wrong with flexibility? (Score:2)
Re:What's wrong with flexibility? (Score:2)
Here in grad school my friends from China never even heard of fortune cookies until they came here. It's like reverse culture-shock.
Re: (Score:2)
Re:What's wrong with flexibility? (Score:2)
Before believing our word, consider using Google to search for "chopstick history". You will discover evidence that leads to the predominant view that chopsticks were in fact invented in China about five thousands years ago.
Having briefly discussed the reliability of the author's facts, let us also comment on matters with regard to the style of the text. There is a natural objection to individuals using grammatical form rather than substance in an attempt to manipulate the reader in p
Re:What's wrong with flexibility? (Score:2)
Re: (Score:2)
OT: Re:Dammit, not another PDF... (Score:2)