I'm Greg DeK. You should all know me by this point, as I've likely contacted each one of you about your interest in an official "Fedora Live CD" project.
Why? Aren't there plenty of Live CDs out there already? Well, yes, there are -- and even some that are based on Fedora. But we think that Fedora-branded Live CDs are still important for a few reasons:
1. Everyone else has one. :)
2. It's an ideal low-risk strategy to allow people to find out what Fedora is all about.
3. It's an ideal place for community contribution and development; the technical barriers are well-understood, and the community has already taken the lead in creating Fedora-based Live CDs. And in this collaboration, we at Red Hat have an opportunity to grow closer to the folks who dig our stuff.
===
So. To get the ball rolling, how about some introductions? I'll start.
I'm Greg DeK, community relations manager for Red Hat. I work on the Fedora Extras project, act as legal go-between for the Fedora project, and do a bunch of other stuff.
Some of my goals for this project:
* Build robust tools for building Fedora-based Live CDs, and putting these tools into Fedora Extras.
* Create Live CDs from these tools, based entirely upon content in Fedora Core and Fedora Extras, for official "blessing".
* Distribute these blessed Live CDs far and wide, through whatever clever means we can come up with.
--g
_____________________ ____________________________________________ Greg DeKoenigsberg ] [ the future masters of technology will have Community Relations ] [ to be lighthearted and intelligent. the Red Hat ] [ machine easily masters the grim and the ] [ dumb. --mcluhan Red Hat Summit ] [ New Orleans ] [ Learn. Network. Experience Open Source. June 1/2/3 2005 ] [ (And Make Your Boss Pay For It.) [ http://www.redhat.com/promo/summit/
On Thursday 31 March 2005 14:25, Greg DeKoenigsberg wrote:
So. To get the ball rolling, how about some introductions? I'll start.
Hello,
I work on SE Linux and the auditing framework for Red Hat. But before I joined Red Hat, I was building a distribution for use in an ISP setting. I had taken FC2.5 and applied 200+ patches and wrote a build system. I was struggling with the idea of how to do the installation - I wasn't a fan of anaconda since its in python and I don't write python code.
I hit upon the idea of distributing the build as a Live CD. This way if someone hacks our machine, we just hit the reset button. Upgrading the machine to a new OS just means ejecting the CD and rebooting to the new media.
I searched around the internet and ran across Basilisk Live CD. I did a little work with Dirk Westfal to get the Live CD creation process documented in a script. This way it is reproducible. I created the Live CD's for that ISP's linux distribution and it all worked.
I have recently integrated the scripts with the Rookery Build system so that people can create Live CD's easily. The Live CD kit can be found at: www.web-insights.net/rookery. You can use pre-built rpms such as Fedora's. I am actively working on the scripts to make them better. If no one else has Live CD creation scripts available, I'd like to offer this as a starting point.
My goals is to help with the technical side and help nail the issues down. That's about the extent of it.
-Steve Grubb
On Thu, 2005-03-31 at 14:54 -0500, Steve Grubb wrote:
I have recently integrated the scripts with the Rookery Build system so that people can create Live CD's easily. The Live CD kit can be found at: www.web-insights.net/rookery. You can use pre-built rpms such as Fedora's. I am actively working on the scripts to make them better. If no one else has Live CD creation scripts available, I'd like to offer this as a starting point.
A while ago I worked on the stateless linux project, and as part of this i wrote the mklivecd script in the readonly-root package. Its pretty simple as live cds go, it just creates a bootable zisofs image from a stateless linux root. The actual work to make a livecd work was part of the standard stateless linux work were everything runs with a readonly filesystem.
I don't currently work on this anymore, but I'll hang around here and see if there is anything i can do to help out.
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= Alexander Larsson Red Hat, Inc alexl@redhat.com alla@lysator.liu.se He's a world-famous flyboy astronaut who hides his scarred face behind a mask. She's a mentally unstable wisecracking detective who dreams of becoming Elvis. They fight crime!
Hi,
On Thu, 2005-03-31 at 14:25 -0500, Greg DeKoenigsberg wrote:
So. To get the ball rolling, how about some introductions? I'll start.
I'm in the slightly strange position of being a Red Hat contractor who does various odd jobs (mostly relating to GFS, since I worked for Sistina in the past). When I'm not working for Red Hat, my main job is running ChyGwyn, better known as The Linux Emporium, a well known supplier of all things Linux related but most famous for supplying various distributions on cheap CDs. As such I'm always keen to see another distribution made available for me to duplicate and sell :-)
Also, this year the UKUUG Linux conference comes to Swansea, UK which is where I'm based. I'm considering doing a tutorial on Linux networking and a Live CD would provide a very useful basis for this I think. I wrote most of the kernel side of the Linux DECnet networking code, so this remains an area of interest for me.
My main constraint is the time I can spend on this project bearing in mind my other commitments.
I am a python programmer (amoung other things) and its the one scripting language which has tempted me away from C and kernel development, but I'm still not so keen on the internals of anaconda :-) I don't think it would be of great help for a Live CD anyway and the initscripts are an area I think would need a number of changes compared with the "standard" Fedora distribution.
One area I'm concerned about is the speed of booting and it would be interesting to see what might be done to address that.
Steve.
On Thu, 2005-03-31 at 14:25 -0500, Greg DeKoenigsberg wrote:
So. To get the ball rolling, how about some introductions? I'll start.
I'm Colin Charles, general all-rounder on The Fedora Project. My interst in making a LiveCD is so that Fedora can get spread far and wide, and that I can also use it for the United Nations LiveCD that I was doing (still am doing). It is currently Ubuntu based because Casper actually "just works"
I didn't want it Stateless based because of the fact that during the pre-FUDCon discussions and just hanging out with most of the @redhat's, they figure Stateless is not really a viable option. Xen/anaconda might be what is its replacement, just minus LiveCDs
So, lets figure it out and get working on it. I've also been talking to Steve Grubb about the Rookery system which I've been meaning to fiddle with
Regards
Colin Charles wrote:
On Thu, 2005-03-31 at 14:25 -0500, Greg DeKoenigsberg wrote:
So. To get the ball rolling, how about some introductions? I'll start.
I'm Colin Charles, general all-rounder on The Fedora Project. My interst in making a LiveCD is so that Fedora can get spread far and wide, and that I can also use it for the United Nations LiveCD that I was doing (still am doing). It is currently Ubuntu based because Casper actually "just works"
I didn't want it Stateless based because of the fact that during the pre-FUDCon discussions and just hanging out with most of the @redhat's, they figure Stateless is not really a viable option. Xen/anaconda might be what is its replacement, just minus LiveCDs
So, lets figure it out and get working on it. I've also been talking to Steve Grubb about the Rookery system which I've been meaning to fiddle with
Regards
Hi, Neville here,
I have some ideas that I would like to see implemented by the Fedora Live CD project. I think you should aim to be at the top of the pack so the Fedora Core live CD should really be a step up from what is currently available. The various Knoppix boot CDs are at the cutting edge and you need to release something far superior to draw people to your products.
As a boot CD builder I have used cloop and squashfs and I would hope that someone would add squashfs to the standard Fedora Core kernel distribution before starting this project. This will allow for the Fedora livecd to have a lot more files.
The ADIOS boot CD, which I have been developing is currently based on Fedora Core 3. ADIOS uses the busybox/uclibc/isolinux startup which then performs a pivot_root to the real filesystem this works fine but I had to modify the filesystem so that /usr was read-only squashfs and the read-write files are all moved into /var which is a RAM drive (or disk partition).
The Xendemo boot CD is a better model, it uses copy-on-write filesystems and I think that including Xen would also make the Fedora livecd very attractive to Educational Institutions, which want to teach networking and security.
I would like to see most of the services that are needed to complete the RedHat Certified Engineer (RHCE) course on the Fedora livecd. And maybe some of the promotional material associated with obtaining the RHCE.
Lastly why not a Fedora live DVD which would support as many languages as possible on startup, as there seems to be a huge non-English following of RedHat Fedora Core. Using squashfs you will have plenty of room to have both the RPMS and the SRPMS on the DVD as well.
----
Myself and Mark Huth are about to start rebuilding the busbox/uclibc software in preparation to building the the next ADIOS boot CD based on Fedora Core 4. We have plans to do most of the above but would be happy for RedHat to do the work, although we could also help with ideas and coding. For comparison you can download the current ADIOS boot CD version 4.11 from http://dc.qut.edu.au/remote/adios4 username: adios password: 12qwaszx Start with run option 1 first, you also need the username adios and password 12qwaszx to login. The FAQ is on the CD but if you don't get that far go to http://dc.qut.edu.au/adios
On Tuesday 05 April 2005 09:46, Neville Richter wrote:
As a boot CD builder I have used cloop and squashfs and I would hope that someone would add squashfs to the standard Fedora Core kernel distribution before starting this project. This will allow for the Fedora livecd to have a lot more files.
You don't have to do it like this. You can use mkzftree during the last stages to create compressed file system that the kernel knows how to read. Reading from the squashfs page, they use gzip so isn't it the same? I looked at cloop and decided there was no gain from mkzftree.
ADIOS uses the busybox/uclibc/isolinux startup which then performs a pivot_root to the real filesystem this works fine but I had to modify the filesystem so that /usr was read-only squashfs and the read-write files are all moved into /var which is a RAM drive (or disk partition).
This is more or less normal for livecds.
The Xendemo boot CD is a better model, it uses copy-on-write filesystems and I think that including Xen would also make the Fedora livecd very attractive to Educational Institutions, which want to teach networking and security.
Right. This is an improvement in terms of memory space. I also include /etc in tmpfs because networking stuff wants to write there.
-Steve Grubb
On Apr 5, 2005 7:00 AM, Steve Grubb sgrubb@redhat.com wrote:
On Tuesday 05 April 2005 09:46, Neville Richter wrote:
As a boot CD builder I have used cloop and squashfs and I would hope that someone would add squashfs to the standard Fedora Core kernel distribution before starting this project.
You don't have to do it like this. You can use mkzftree during the last stages to create compressed file system that the kernel knows how to read.
That's an interesting approach that I thought about using on another project. As a comparison I built the same CD using both zisofs and cloop. IIRC, cloop ended up providing 10 to 15 percent better compression. Unfortunately cloop has it's problems too. One of them has been keeping the Debian oriented sources compiling against the Fedora kernels.
I like Neville's suggestion of having squashfs support in the kernel. This would provide both the convenience of zisofs and the compression advantages of cloop.
-Tom
Tom Lisjac wrote:
On Apr 5, 2005 7:00 AM, Steve Grubb sgrubb@redhat.com wrote:
On Tuesday 05 April 2005 09:46, Neville Richter wrote:
As a boot CD builder I have used cloop and squashfs and I would hope that someone would add squashfs to the standard Fedora Core kernel distribution before starting this project.
You don't have to do it like this. You can use mkzftree during the last stages to create compressed file system that the kernel knows how to read.
That's an interesting approach that I thought about using on another project. As a comparison I built the same CD using both zisofs and cloop. IIRC, cloop ended up providing 10 to 15 percent better compression. Unfortunately cloop has it's problems too. One of them has been keeping the Debian oriented sources compiling against the Fedora kernels.
I like Neville's suggestion of having squashfs support in the kernel. This would provide both the convenience of zisofs and the compression advantages of cloop.
-Tom
-- Fedora-livecd-list mailing list Fedora-livecd-list@redhat.com https://www.redhat.com/mailman/listinfo/fedora-livecd-list
Hello again,
I would hope that the COW filesystem technology used on the Xendemo CD becomes the standard way of starting the Fedora live-CD.
My current problem which will also be a problem for the Fedora live-CD is that Schools and Universities are all moving towards wireless laboratories. The problem is that wireless vendors seem to cater primarily for MS Windows and you then have to use software that runs these DLLs. Also the licence agreements for the wireless software do not extend to distributing vendor software with boot CD.
On Tue, 2005-04-05 at 18:42 -0600, Tom Lisjac wrote:
On Apr 5, 2005 7:00 AM, Steve Grubb sgrubb@redhat.com wrote:
On Tuesday 05 April 2005 09:46, Neville Richter wrote:
As a boot CD builder I have used cloop and squashfs and I would hope that someone would add squashfs to the standard Fedora Core kernel distribution before starting this project.
You don't have to do it like this. You can use mkzftree during the last stages to create compressed file system that the kernel knows how to read.
That's an interesting approach that I thought about using on another project. As a comparison I built the same CD using both zisofs and cloop. IIRC, cloop ended up providing 10 to 15 percent better compression. Unfortunately cloop has it's problems too. One of them has been keeping the Debian oriented sources compiling against the Fedora kernels.
I like Neville's suggestion of having squashfs support in the kernel. This would provide both the convenience of zisofs and the compression advantages of cloop.
The best way to get this done is to try to get squashfs into the upstream kernel. There has been some motion on this recently:
http://www.ussg.iu.edu/hypermail/linux/kernel/0503.1/2040.html
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= Alexander Larsson Red Hat, Inc alexl@redhat.com alla@lysator.liu.se He's an unconventional Catholic astronaut possessed of the uncanny powers of an insect. She's a man-hating communist femme fatale with a birthmark shaped like Liberty's torch. They fight crime!
livecd@lists.fedoraproject.org