Plan 9 from Bell Labs comes from the same stable as the UNIX operating system, which of course Linux was designed after, and Apple’s OS X runs on top of a certified UNIX operating system. Just like UNIX, Plan 9 was developed as a research OS — a vehicle for trying out new concepts — with it building on key UNIX principles and taking the idea of devices are just files even further.
In this post, we take a quick look at the Plan 9 OS and some of the notable features, before moving on to the construction of a self-contained 4-node Raspberry Pi cluster that will provide a compact platform for experimentation.
The post is almost a year old, but it hasn’t ever appeared here, and for that I will not stand. Plan 9 gets little attention and press, and that’s decidedly a shame.
Shame they don’t go into any details. Network diagram? Which RaspPi does which role?
I couldn’t find the download for RB PI 3B+? I only had dead links.
During 2018 Nokia pulled the plug from cs.bell-labs.com which hosted plan9.bell-labs.com.
https://www.mail-archive.com/[email protected]/msg36836.html
https://www.mail-archive.com/[email protected]/msg37681.html
The article has dead links and correct links. The one you’re looking for has at least these two mirrors:
https://9p.io/sources/contrib/miller/9pi.img.gz
https://plan9.io/sources/contrib/miller/9pi.img.gz
You can browse the sources (system and contributions) from:
https://plan9.io/sources/
Looks like the url changed?
https://9p.io/plan9/
https://9p.io/sources/contrib/miller/ – for the RPi Image
I’m considering to replace SMB with 9p for my personal Linux-based fileserver; my clients at home are also running Linux (mostly). I never really liked NFS authentification schemes for my simple needs, and sshfs is somewhat lacking (but works).
It would be great to have more network transparency. But, AFAIK, the most useful Plan 9 ideas are already adopted and/or sidelined by newer Linux / GNU stuff. Only if Wayland would have network transparency… I’m sticking to old X11 for the time being.
https://wiki.gnome.org/Projects/Mutter/RemoteDesktop
Wayland support is there now, but has some severe limitations (encryption and authentication, specifically). But its finally coming.
I was looking for network transparency. Not for a remote desktop protocol.
Out of curiosity, what is your usecase? While doing odd things like having the ui of apps render locally while the application ran remotely was interesting, I never found a great usecase. In most cases you might as well use a remote desktop protocol.
Very often I ssh into a box. Sometimes I need a graphical tool. Then I just start the tool. X11 forwarding is enabled by default for my machines. It works very fast (local LAN) and is very convenient.
I do not have any need for some “tool”. It already works, using ssh. No need to type more than, e.g. “virt-manager” to start up the GUI tool to manage my virtual machines on my server.
I have no need to discuss “use cases”; I want network transparency. Like the one already existing and working.
Why is that so hard to understand?
evert,
Honestly, dismissing the importance of use cases isn’t helping your cause. Bill Shooter of Bul is a smart guy and he was obviously asking for use cases to understand why remote desktop isn’t good enough for you.
I actually agree with you that remote X11 displays are awesome. People who’ve only done full screen RDP may not realize how much nicer it is to manage remote applications under the local desktop manager. When I was at school (technically off campus accessing lab computers remotely) X11 was very helpful since remote apps and local apps could effectively run side by side. We could do our work with ssh and when we needed a GUI program, we just launched it instantly from there, no need to log into a new GUI session only to have to find our place again.
I don’t use remote X11 GUIs any more, but there are still times I’d find them useful. An example is that the best system resource monitors (networking/memory/disk/etc) are GUI based interfaces that offer significantly better representations than text interfaces (ie convey more information using less screen real estate). It would be very handy at times to launch one of these GUIs apps on remote servers and have them display on my local machine in a small window. Launching a full blown desktop session is undesirable and just gets in the way.
Cool, but it’s not a cluster is it? He booted into one of them – am I missing something about Plan 9? Does it automatically distribute itself to everything on the cluster? If so, it’d be cool to see some sort of diagnostics or stats of a running task and how it’s distributed.
“Honestly, dismissing the importance of use cases isn’t helping your cause. Bill Shooter of Bul is a smart guy and he was obviously asking for use cases to understand why remote desktop isn’t good enough for you.”
Bill Shooter is an idiot. If he doesn’t know already know enough about remote X11 displays to understand why remote desktops aren’t a good replacement for them, how is asking/demanding for some moronic “usecase” that remote desktops aren’t a good fit for people to use going to solve anything in this instance?
It’s like watching a dog trying to catch his tail.
…and you’re back to your “normal” self.