[katello-devel] Fedora 18 state

Fri Feb 22 14:40:21 UTC 2013

Ok this looks good then. 

Then it looks like a viable alternative. Thanks for the update.

LZ

On Fri, Feb 22, 2013 at 09:34:36AM -0500, David Davis wrote:
> So a lot of the improvements from REE are built into MRI 1.9.3. 1.9.3 has a lot of the same configuration options like RUBY_GC_MALLOC_LIMIT, RUBY_HEAP_MIN_SLOTS, and RUBY_FREE_MIN. So there's probably not going to be a version of REE for 1.9.3.
> 
> I probably would not recommend changing to Passenger because of one error but I think it would be nice to consider and test other platforms.
> 
> David
> 
> ----- Original Message -----
> > From: "Lukas Zapletal" <lzap at redhat.com>
> > To: katello-devel at redhat.com
> > Sent: Friday, February 22, 2013 9:24:34 AM
> > Subject: Re: [katello-devel] Fedora 18 state
> > 
> > Yes, passenger is the number one for Rails. Thin can scale really
> > well,
> > but mainly for simple apps with less I/O operations, best to use with
> > async (evented) apps. We (and Rails in general) are not apparently
> > good
> > fit for Thin.
> > 
> > But I wonder how passenger works on non-enterprise ruby edition like
> > MRI
> > Ruby. This is not copy-on-write friendly like REE and I am not sure
> > if
> > REE is in Fedora (what license it is by the way - not possible to
> > google
> > on the official pages, they are outdated, invalid links, outdated
> > information).
> > 
> > I wonder if this effort is dead or not (REE does not support Ruby 1.9
> > AFAIK) and how does it perform with MRI Ruby, which we will
> > apparently
> > use.
> > 
> > So I am afraid, Katello is not the Ordinary Ruby Production (TM)
> > application where you just upload your rb files to a provider which
> > is
> > running unknown-licence highly-custom REE 1.8 with passenger. We have
> > our own platform which is based on MRI with Fedora/RHEL patches.
> > 
> > These are my main concerns. I am not against passanger at all, we
> > just
> > need to talk about this more, maybe some proof of concept with some
> > tests. But it will not be like what the passenger page says:
> > 
> > "33% memory reduction and faster? Is this for real?"
> > 
> > Definitely against "hey there is a bug, let's move to passenger" (no
> > offence at all - it is surely a valid option :-)
> > 
> > LZ
> > 
> > 
> > On Fri, Feb 22, 2013 at 08:57:40AM -0500, David Davis wrote:
> > > Most people are using passenger in production [1]. Although other
> > > options like unicorn and torquebox are becoming more popular.
> > > 
> > > Regarding scaling, I think it would be easy to set up katello
> > > running in passenger and then run siege against it.
> > > 
> > > 1
> > > https://planetargon-blog.s3.amazonaws.com/images/infographic-2012-survey.png
> > > 
> > > David
> > > 
> > > ----- Original Message -----
> > > > From: "Bryan Kearney" <bkearney at redhat.com>
> > > > To: katello-devel at redhat.com
> > > > Sent: Friday, February 22, 2013 8:46:02 AM
> > > > Subject: Re: [katello-devel] Fedora 18 state
> > > > 
> > > > On 02/22/2013 08:39 AM, Lukas Zapletal wrote:
> > > > > After studying possible deployment solutions for Ruby, I am not
> > > > > under
> > > > > impression mod_passenger is the best technology out there and I
> > > > > am
> > > > > afraid we will avoid this the that, but we can expect other
> > > > > issues.
> > > > >
> > > > > I would like to talk about this more, maybe after my
> > > > > presentation
> > > > > next
> > > > > deep dive. I wonder how will this scale up, because passenger
> > > > > is
> > > > > strictly forking only solution and katello boot time is about
> > > > > 15
> > > > > seconds. It can work well with REE/Rubinius patched Rubies with
> > > > > better
> > > > > forking memory management, but I wonder what this will do with
> > > > > MRI
> > > > > Ruby
> > > > > and Katello.
> > > > >
> > > > > My major concern is this is out of the frying pan into the
> > > > > fire.
> > > > > I'd
> > > > > rather postpone mod_passenger talk a little bit more.
> > > > >
> > > > > Isn't there really a workaround?
> > > > 
> > > > 
> > > > I thought that mod passenger was the defacto rails deployment
> > > > model
> > > > right now? Is each request a new process? I thought there was
> > > > pooling
> > > > going on.
> > > > 
> > > > -- bk
> > > > 
> > > > 
> > > > _______________________________________________
> > > > katello-devel mailing list
> > > > katello-devel at redhat.com
> > > > https://www.redhat.com/mailman/listinfo/katello-devel
> > > > 
> > > 
> > > _______________________________________________
> > > katello-devel mailing list
> > > katello-devel at redhat.com
> > > https://www.redhat.com/mailman/listinfo/katello-devel
> > 
> > --
> > Later,
> > 
> >  Lukas "lzap" Zapletal
> >  #katello #systemengine
> > 
> > _______________________________________________
> > katello-devel mailing list
> > katello-devel at redhat.com
> > https://www.redhat.com/mailman/listinfo/katello-devel
> > 

-- 
Later,

 Lukas "lzap" Zapletal
 #katello #systemengine