Fedora Core May Be Reborn

darthcamaro writes "At the first ever Fedora Flock conference this past weekend, a proposal was put forward by developer Mat Miller to re-architect Fedora with a core distribution, surrounded by layers of additional functionality for desktop, server and cloud. It's a proposal that Fedora Project Leader Robyn Bergeron is interested in too. 'How can we make Fedora be something that is modular enough to fit into all those different environments (device, desktop, server & cloud) , while still acknowledging that a one-size-fits-all approach isn't something that draws people into the project?' Bergeron said. 'People want something that is specifically for them.'"

Comments

Arch Linux

[mfwitten]

Hasn't Arch Linux already solved this problem (for at least x86)?

Hasn't Gentoo already solved this problem for [almost] all architectures?

Re:Arch Linux

[Anonymous Coward]

But you wouldn't use those in an office/cloud environment, which is what I think the developers are thinking about.

i wouldnt? whoops...my bad. guess i should redo all the servers I have been running for the past 6 years

Gentoo and Arch don't have recognised large-scale support vendors, which is something many companies require. I would guess any work will be back-ported into a RedHat if something fruitful comes from this.

Re:Arch Linux

[mfwitten]

Over the years that I've used these systems, that hasn't been my experience. So, readers: YMMV.

In fact, mileage often depends on whose in the driver's seat.

I use CentOS for those unless I have no choice

[raymorris]

Fedora isn't really targeted to those environments anyway.
That's where you want stability, well proven packages and long term maintenance.
Fedora is the cutting edge, better suited to an enthusiast desktop or maybe a development environment.

In the past I used Fedora for office and servers. That was an error. Switching to CentOS (which is Fedora stabilized) was a much better decision.

The exception was one case in which I needed a brand new subsystem - kernel plus userspace. For that, Fedora made sense because the brand new version I needed was not on RedHat / CentOS yet.

Re:I use CentOS for those unless I have no choice

[Peter H.S.]

I think a lot of the traditional thinking about servers are changing: VM's, OS-containers, cloud computing means, that the traditional big iron, unchanging OS with hand crafted config files, is receding. Instead all the interesting stuff will happen in mass deployed auto configured VM's, or OS-containers (OpenVZ, LXC etc). There will be VPS's (Virtual Private Servers, served from the cloud), etc. etc.

With services doing live migrations across data centers, and services being configured and executed on a demand only basis, and everything interesting in fleeting VM's and containers, they need for long term feature stable servers may recede;

So while Fedora may not be everyone's choice for a bare metal VM and container server, it may very well be exactly what you want to put inside those VM's and containers. This is why it is important to rethink the Fedora distro. As it is now, its base install very much reflect old style UNIX thinking (nothing wrong with UNIX style, but such servers are not the whole story), that means the base install pulls in stuff like "ed", "tar", "file" etc. While they may nice little standard programs, they may not make sense in a custom VPS.

Re:What Fedora Needs...

[techno-vampire]

Fedora is all about being a testbed for new programs, new technology, new ideas. It's not about being stable, it never has been and it probably never will be. If you want a RedHat based distro with community support that's stable, you don't want Fedora, you want CentOS.

Really? Give it a break.

[adosch]

Man, I am so sick of this 're-birth' crap from Fedora. I liked Fedora 'core' back 7+ years ago before we had to be this uber bleeding edge -slash- agile uber aggressive build cycle that fucks everything up and obsoletes distribution usage to about 6 months.

When it was 'just' an upstream snapshot look to what RedHat Enterprise was going to be in the future, I was totally cool with that, and it melded nicely in a lot of environments. But that spin-off has become such a damn mess now with developer heavy ideas that, in some case, go against every foundation of a traditional UNIX-like operating system design, I could really give who shits what the do now.

Making a 'one-size-fits-all' OS is, pain and simple: a horrible idea. I don't want a damn highly integrated OS that I can use for everything. You'll never get that right, and some 'next-in-line' guy they give 5 minutes of talk time at the next conference will say the same thing.

When you take shit, and try and re-invent it with only shit, I'm sure everyone knows the result you get.

Re:Fedora is very important

[lister king of smeg]

With integrated secure boot and remote attestation Fedora is likely to be the *only* Linux distribution which can lawfully be used on the Internet in the coming years. It is important that they continue their good work if we are to have any Linux at all.

ah I believe ubuntu has its bootloader signed and did not the linux foundation have shim loader signed monthes ago?

They should start by not breaking things

[Anonymous Coward]

Well i like GNOME 3 but the subsequent changes are way too much:

1. no dock, preferrably configurable one (left or right)

2. removing F3 from Nautilus

3. changing gvfs paths every 2nd relaease

4. moving menu options into the main title bar (like preferences in GNOME)

I've been using Fedora since the beginning but after F19 I am no longer recommending it to anyone.

Start with a LTS distro

[marcovje]

Since 18 months updates is simply not enough.

Re:Really? Give it a break.

[Karrde712]

I run Fedora 19 with KDE on my home machines. Basically I set aside about 5 to 6 hours every six months to upgrade and when I say upgrade I mean a complete re-installation of the latest Fedora from DVD. Even though I actually backup my data (over 1TB) progressively I have never had to recover my data since I use LVM to manage my disks and all I have to do is make sure the system volumes including swap and of course the /boot partition are sized properly. Actually the only time I did have to recover my data when I converted the file-system from ext3 to ext4 and IMHO the performance improvement was worth it.

I strongly recommend that you try upgrading with 'fedup' next time around. It's far-and-away better than our historical upgrade processes and works in-place. I've personally gone from F17->F18->F19 using it with no ill effects.

I have actually found Fedora from 10 onward have been remarkably stable although I will admit when KDE 4.0 came out (I think that was with Fedora 15) I actually switched to Gnome util they fixed the stability issues, however that was not a Fedora issue.

Would I recommend Fedora for the Enterprise? Hell no! since you want any enterprise solution to be supported and in large corporations this usually means a Microsoft OS (this is changing but slowly) for the desktop and a mix of Linux (in my experience Redhat), Microsoft and Unix for the server room.

That's going to depend on your definition of Enterprise. Would I recommend Fedora today as your long-term FreeIPA or other core infrastructure server? No, probably not. On the other hand, would I recommend it for DevOps and rapidly deployed-used-and-killed VM instances for newer technologies such as Ruby on Rails or Node.js? Absolutely. Fedora's rapid development cycle is much more in line with those DevOps behaviors. It's actually a myth that "Fedora isn't for production". I know a great many DevOps deployments using Fedora successfully.

That all said, the major piece that was missing from this incredibly (and clearly intentionally) misleading summary is that the purpose of splitting off Fedora into three targets is to provide better support for those who want to use Fedora in production (the cloud image), those who want to develop their layered software so that it will run on the next version of RHEL/CentOS (the server) and people who want a comprehensive desktop for getting stuff done (the workstation/client).

Stephen Gallagher (Fedora Engineering Steering Committee)