Moving CFS forward

[Jonathan "Duke" Leto]

Howdy Y'all,

As a suggestion for moving forward, perhaps we should ask pblittle for
his branch that fixes issue 183 of vcap:

https://github.com/cloudfoundry/vcap/issues/183#issuecomment-5578177

He is a pretty friendly guy, so I have cc'ed him.

Duke

PS: I know I have been incommunicado lately. Buying a house, moving
and getting married in the same month
will do that to you. I will be back soon :)

--
Jonathan "Duke" Leto <jona...@leto.net>
Leto Labs LLC http://labs.leto.net
209.691.DUKE http://dukeleto.pl

[M. Edward (Ed) Borasky]

On Tue, May 8, 2012 at 10:40 AM, Jonathan "Duke" Leto <jona...@leto.net> wrote:
> Howdy Y'all,
>
> As a suggestion for moving forward, perhaps we should ask pblittle for
> his branch that fixes issue 183 of vcap:
>
> https://github.com/cloudfoundry/vcap/issues/183#issuecomment-5578177
>
> He is a pretty friendly guy, so I have cc'ed him.
>
> Duke
>
> PS: I know I have been incommunicado lately. Buying a house, moving
> and getting married in the same month
> will do that to you. I will be back soon :)

Ah ... but how about the rest of us? Everyone, myself included, seems
to have re-prioritized away from CloudFreeStyle. I'm releasing
Computational Journalism Server 1.2.5 in a day or so and this will
probably be a feature freeze as far as openSUSE, openSUSE Build
Service and SUSE Studio is concerned.

If you'll pardon the pun, I think the desktop is greener on Fedora and
Ubuntu for both OpenStack IaaS and open source PaaS (OpenShift Origin
and CloudFreeStyle). The disk failure on the openSUSE Build Service
main download server last week didn't make things any better, but that
was more an annoyance than a show-stopper like terrible / non-existent
documentation or a b0rked community governance process. ;-)

--
Twitter: http://twitter.com/znmeb Computational Journalism Server
http://j.mp/compjournoserver

Data is the new coal - abundant, dirty and difficult to mine.

[Jonathan "Duke" Leto]

Howdy,

> Ah ... but how about the rest of us? Everyone, myself included, seems
> to have re-prioritized away from CloudFreeStyle.

Huh? I didn't get that impression at all.

Duke

[Ingy dot Net]

Please know that I am actively scheming on CFS. Stay tuned...

Ingy

[M. Edward (Ed) Borasky]

On Tue, May 8, 2012 at 1:11 PM, Ingy dot Net <in...@ingy.net> wrote:
> Please know that I am actively scheming on CFS. Stay tuned...
>
> Ingy

Excellent!

>
> On May 8, 2012 12:28 PM, "Jonathan "Duke" Leto" <jona...@leto.net> wrote:
>>
>> Howdy,
>>
>> > Ah ... but how about the rest of us? Everyone, myself included, seems
>> > to have re-prioritized away from CloudFreeStyle.
>>
>> Huh? I didn't get that impression at all.
>>
>> Duke
>>
>>
>> --
>> Jonathan "Duke" Leto <jona...@leto.net>
>> Leto Labs LLC http://labs.leto.net
>> 209.691.DUKE http://dukeleto.pl

[P. Barrett Little]

Hi folks,

I apologize for the delayed response, life is a little hectic around
here with a relatively new baby. :)

I'll be glad to submit a pull request for this feature. Please give me
a few days to make the request.

Concerning the current state of the CFS fork, what is the goal of
keeping it in sync with the CF master branch? The CF master is moving
very rapidly these days.

Thanks,
Barrett

On May 8, 1:40 pm, "Jonathan \"Duke\" Leto" <jonat...@leto.net> wrote:
> Howdy Y'all,
>
> As a suggestion for moving forward, perhaps we should ask pblittle for
> his branch that fixes issue 183 of vcap:
>
> https://github.com/cloudfoundry/vcap/issues/183#issuecomment-5578177
>
> He is a pretty friendly guy, so I have cc'ed him.
>
> Duke
>
> PS: I know I have been incommunicado lately. Buying a house, moving
> and getting married in the same month
> will do that to you. I will be back soon :)
>
> --

> Jonathan "Duke" Leto <jonat...@leto.net>
> Leto Labs LLChttp://labs.leto.net
> 209.691.DUKEhttp://dukeleto.pl

[Ingy dot Net]

On Thu, May 10, 2012 at 10:41 AM, P. Barrett Little <bar...@barrettlittle.com> wrote:

Hi folks,

I apologize for the delayed response, life is a little hectic around
here with a relatively new baby. :)

I'll be glad to submit a pull request for this feature. Please give me
a few days to make the request.

Concerning the current state of the CFS fork, what is the goal of
keeping it in sync with the CF master branch? The CF master is moving
very rapidly these days.

This is the $20 question. We need to figure out what the goals of CFS are. I think that depends a lot on whether we get significant code contributions back from various closed forks, which will take some time to sort out.

I think the safe thing to do while that is being figured out is to keep master in sync with CF. We can develop on branches and let things go in a few directions until clarity comes.

[M. Edward (Ed) Borasky]

On Thu, May 10, 2012 at 1:31 PM, Ingy dot Net <in...@ingy.net> wrote:
> On Thu, May 10, 2012 at 10:41 AM, P. Barrett Little
> <bar...@barrettlittle.com> wrote:
>>
>> Hi folks,
>>
>> I apologize for the delayed response, life is a little hectic around
>> here with a relatively new baby. :)
>>
>> I'll be glad to submit a pull request for this feature. Please give me
>> a few days to make the request.
>>
>> Concerning the current state of the CFS fork, what is the goal of
>> keeping it in sync with the CF master branch? The CF master is moving
>> very rapidly these days.
>
>
> This is the $20 question. We need to figure out what the goals of CFS are. I
> think that depends a lot on whether we get significant code contributions
> back from various closed forks, which will take some time to sort out.
>
> I think the safe thing to do while that is being figured out is to keep
> master in sync with CF. We can develop on branches and let things go in a
> few directions until clarity comes.

How many other forks are there? I haven't looked at the "official"
Cloud Foundry repositories recently, but I'm assuming there's at least
AppFog, Stackato and PaaS.io in addition to CloudFreeStyle.

[Jonathan "Duke" Leto]

Howdy,

This email happened just before my wedding, so excuse the lag :)

Basically, every company that builds on CF maintains a fork of it so
they can actually run it in production. These are "tiny forks" that
usually get rebased/merged with newer CF code. This number is probably
around 10 or 20, but it moves fast and I could be off on that.

A larger fork is IronFoundry, which is the .NET fork of CF.

As far as I can tell, VMware ignored all requests to merge code from
IronFoundry, so that is a "hard fork." That being said, ActiveState +
IronFoundry seem to be working together, so it is hard to say exactly
where everything stands now.

Duke

[Adron Hall]

Also note, CF also has splintered the project into multiple repositories, or plans to very shortly (I haven't look at it in a few weeks, instead focusing on other CF + IF concerns).

Has anyone else been diving into the main CF to check its status?

--

Adron B Hall

Tech Blog: http://compositecode.com

Iron Foundry Project: http://www.ironfoundry.org

About Me: http://compositecode.com/about

Twitter: http://www.twitter.com/adron

[M. Edward (Ed) Borasky]

On Wed, Jun 6, 2012 at 3:03 PM, Adron Hall <adro...@gmail.com> wrote:
> Also note, CF also has splintered the project into multiple repositories, or
> plans to very shortly (I haven't look at it in a few weeks, instead focusing
> on other CF + IF concerns).
>
> Has anyone else been diving into the main CF to check its status?

I haven't touched CFS or Cloud Foundry since about a week after the
hackathon. I've got what I think is the near-final release of my
openSUSE appliance done and am ready to move on to making a PaaS
appliance. However, I strongly suspect I'll have an OpenShift version
first simply because the documentation for making a "do-it-yourself"
version is pretty good already and Red Hat is bending over backwards
helping developers in their forums.

At this stage of my project development, I'm looking for stuff that
works *now* or can be made to work with just a few lines of code. I'm
*not* looking for a project to rename an existing project or document
an existing tool set, and I'm most assuredly *not* interested in
learning a new programming language! The only viable options for me at
this point are to do everything in ActivePerl / PostgreSQL on Stackato
or build in R from the ground up on OpenShift using whatever glue
logic OpenShift provides.

[Jonathan "Duke" Leto]

Howdy,

> At this stage of my project development, I'm looking for stuff that
> works *now* or can be made to work with just a few lines of code. I'm
> *not* looking for a project to rename an existing project or document
> an existing tool set, and I'm most assuredly *not* interested in
> learning a new programming language!

I hear ya! I think I can safely say that most people looking into PAAS these
days are in the same boat. They don't want to rehaul the engine, they just want
something that works.

> The only viable options for me at
> this point are to do everything in ActivePerl / PostgreSQL on Stackato
> or build in R from the ground up on OpenShift using whatever glue
> logic OpenShift provides.

I am very interested in learning more about OpenShift by hacking on the
insides, so please let me know if I can help you get R + OpenShift on speaking
terms.

So far, the main place where CF beats OpenShift currently (IMHO) is the use of
Linux containers to isolate apps (Warden). Currently OpenShift uses the
standard cgroups and OS-level permissions.

I did talk to them about this on IRC and they seem open to adding containers if
the community really wants it.

Duke

[M. Edward (Ed) Borasky]

On Fri, Jun 8, 2012 at 10:25 AM, Jonathan "Duke" Leto <jona...@leto.net> wrote:
> Howdy,
>
>> At this stage of my project development, I'm looking for stuff that
>> works *now* or can be made to work with just a few lines of code. I'm
>> *not* looking for a project to rename an existing project or document
>> an existing tool set, and I'm most assuredly *not* interested in
>> learning a new programming language!
>
> I hear ya! I think I can safely say that most people looking into PAAS these
> days are in the same boat. They don't want to rehaul the engine, they just want
> something that works.
>
>> The only viable options for me at
>> this point are to do everything in ActivePerl / PostgreSQL on Stackato
>> or build in R from the ground up on OpenShift using whatever glue
>> logic OpenShift provides.
>
> I am very interested in learning more about OpenShift by hacking on the
> insides, so please let me know if I can help you get R + OpenShift on speaking
> terms.

At this point I think the documentation is sufficient ... it's the
proxy stuff I need to research, not what's happening inside OpenShift.

>
> So far, the main place where CF beats OpenShift currently (IMHO) is the use of
> Linux containers to isolate apps (Warden). Currently OpenShift uses the
> standard cgroups and OS-level permissions.

They also have different mechanisms in the OpenShift Origin
("micro-cloud") and OpenShift (Red Hat-hosted) platforms for a few
things. There's some engineering required to make that go away /
integrate the mechanisms. It may be an IP / legal issue too, not just
engineering. There were some pieces that didn't get released open
source.

> I did talk to them about this on IRC and they seem open to adding containers if
> the community really wants it.

They're counting on the community to build quite a bit of that. Still,
as long as all the OS-level operations are the *same* Linux - Ubuntu
in CF or RHEL / Fedora in OpenShift - why does there need to be the
"extra layer" of LXC? Aren't cgroups "the simplest thing that works?"

My first impression from looking at the projects they have on Github
was that every major open source LAMP stack gizmo - WordPress,
MediaWiki, Piwik, etc. - was done, even though I can't imagine why
anyone would run a WordPress blog or MediaWiki wiki anywhere but on a
commercial hosting service. IMHO they should be looking for a "killer
app", not re-inventing LAMP stack hosting. ;-)

[M. Edward (Ed) Borasky]

Speaking of R in the cloud, I found a *very* interesting project. It's
called OpenCPU (http://public.opencpu.org/pages/) and the license is
AGPL 3 "Affero" (http://www.gnu.org/licenses/agpl-3.0.html). It's
built on Ubuntu Oneiric Ocelot but I suspect backporting to
CloudFreeStyle's Ubuntu won't be too difficult. I'm going to see
what's involved in porting to OpenShift Origin, although I'd really
prefer RStudio as an interface over OpenCPU amd RStudio Server already
works on both RHEL and Ubuntu