Carl Youngblood
2005-Jul-01 16:03 UTC
Railsday -- Demonstrating that Rails "can''t scale"?
Now, I know that Rails can actually scale, and I very much appreciate all those who are donating their time, energy and resources to make Railsday possible, but has anyone considered the negative publicity that _will_ occur if the first and third place sites don''t even run? If Railsday is being run on a dedicated server I can''t imagine why it would be incapable of handling the load. I don''t think that it is getting anywhere near a slashdot-sized crowd. Another annoyance is that even if we try to download the site and run it ourselves, just to see what it is like, we are extremely hampered, because it is only available in an online svn browser. Please at least put up tarballs of each site for us to download. Thanks, Carl _______________________________________________ Rails mailing list Rails-1W37MKcQCpIf0INCOvqR/iCwEArCW2h5@public.gmane.org http://lists.rubyonrails.org/mailman/listinfo/rails
Ben Bleything
2005-Jul-01 16:09 UTC
Re: Railsday -- Demonstrating that Rails "can''t scale"?
On Fri, Jul 01, 2005, Carl Youngblood wrote:> Now, I know that Rails can actually scale, and I very much appreciate all > those who are donating their time, energy and resources to make Railsday > possible, but has anyone considered the negative publicity that _will_ occur > if the first and third place sites don''t even run?I don''t think it''s a load issue, I think the server config just got tweaked. They ran fine at one point and then just magically stopped. The one remaining app is still very snappy, so it really doesn''t seem like load.> If Railsday is being run on a dedicated server I can''t imagine why it would > be incapable of handling the load. I don''t think that it is getting anywhere > near a slashdot-sized crowd.I''m sure the railsday server took a hell of a lot more load during the actual competition than it does now. Over a hundred groups using the database and web servers? That''s probably more than 3 apps now.> Another annoyance is that even if we try to download the site and run it > ourselves, just to see what it is like, we are extremely hampered, because > it is only available in an online svn browser. Please at least put up > tarballs of each site for us to download.I''m not sure about this either. How many people are going to have interest in setting up a Rails app but not know how to check something out of subversion? It''s really up to the authors of the projects to provide that service anyway. Remember, the code doesn''t belong to Railsday. I for one am continuing development on my app and I know many others are too. Ben
Sean T Allen
2005-Jul-01 16:11 UTC
Re: Railsday -- Demonstrating that Rails "can''t scale"?
Carl Youngblood wrote:> Now, I know that Rails can actually scale, and I very much appreciate > all those who are donating their time, energy and resources to make > Railsday possible, but has anyone considered the negative publicity > that _will_ occur if the first and third place sites don''t even run?+1 it certainly doesnt make for good publicity. _______________________________________________ Rails mailing list Rails-1W37MKcQCpIf0INCOvqR/iCwEArCW2h5@public.gmane.org http://lists.rubyonrails.org/mailman/listinfo/rails
Carl Youngblood
2005-Jul-01 16:17 UTC
Re: Railsday -- Demonstrating that Rails "can''t scale"?
Thanks, actually, I didn''t realize that you can pass these URLs directly to svn. I''ve been using svn exclusively in svn+ssh mode and am not really familiar with all of its features. On 7/1/05, Ben Bleything <ben-TGHtUsa5cOzMFIMGWPqnnw@public.gmane.org> wrote:> > I''m not sure about this either. How many people are going to have > interest in setting up a Rails app but not know how to check something > out of subversion? >_______________________________________________ Rails mailing list Rails-1W37MKcQCpIf0INCOvqR/iCwEArCW2h5@public.gmane.org http://lists.rubyonrails.org/mailman/listinfo/rails
Ben Bleything
2005-Jul-01 16:22 UTC
Re: Railsday -- Demonstrating that Rails "can''t scale"?
On Fri, Jul 01, 2005, Sean T Allen wrote:> >I''m sure the railsday server took a hell of a lot more load during the > >actual competition than it does now. Over a hundred groups using the > >database and web servers? That''s probably more than 3 apps now. > > > > > I doubt its a scaling issue either but it still doesnt look good when a site > geared towards a demonstration of capability returns errors left and right.Oh yeah, I absolutely agree with that. It''s just bad to blame it on scaling if that''s not the actual issue. People read these lists as much as they go to websites :) Ben
Sean T Allen
2005-Jul-01 16:22 UTC
Re: Railsday -- Demonstrating that Rails "can''t scale"?
Ben Bleything wrote:>On Fri, Jul 01, 2005, Carl Youngblood wrote: > > >>Now, I know that Rails can actually scale, and I very much appreciate all >>those who are donating their time, energy and resources to make Railsday >>possible, but has anyone considered the negative publicity that _will_ occur >>if the first and third place sites don''t even run? >> >> > >I don''t think it''s a load issue, I think the server config just got >tweaked. They ran fine at one point and then just magically stopped. >The one remaining app is still very snappy, so it really doesn''t seem >like load. > > > >>If Railsday is being run on a dedicated server I can''t imagine why it would >>be incapable of handling the load. I don''t think that it is getting anywhere >>near a slashdot-sized crowd. >> >> > >I''m sure the railsday server took a hell of a lot more load during the >actual competition than it does now. Over a hundred groups using the >database and web servers? That''s probably more than 3 apps now. > >I doubt its a scaling issue either but it still doesnt look good when a site geared towards a demonstration of capability returns errors left and right. _______________________________________________ Rails mailing list Rails-1W37MKcQCpIf0INCOvqR/iCwEArCW2h5@public.gmane.org http://lists.rubyonrails.org/mailman/listinfo/rails
François Beausoleil
2005-Jul-01 16:36 UTC
Re: Railsday -- Demonstrating that Rails "can''t scale"?
Carl Youngblood said the following on 2005-07-01 12:17:> Thanks, actually, I didn''t realize that you can pass these URLs directly > to svn. I''ve been using svn exclusively in svn+ssh mode and am not > really familiar with all of its features.For reference, yours and everyone else''s, here''s a link to the Subversion book: http://svnbook.red-bean.com/ Enjoy ! François
John Carlin
2005-Jul-01 17:37 UTC
Re: Railsday -- Demonstrating that Rails "can''t scale"?
Apologies in advance if I''m all over the board on this reply. The general theme is "learning process" and the disclaimer is that I am one of the runners-up (Whack-a-Poll). First off, congrats to the winners! I have a sneaking suspicion that you''re reading the list.Next, thanks to the whole Rails Day team for taking the time to organize and manage this first-ever event. I could only imagine the scope of effort required to pull it off as nothing short of ginormous. OK.. my take on things... On 7/1/05, Carl Youngblood <carl.youngblood-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> wrote:> possible, but has anyone considered the negative publicity that _will_ occur > if the first and third place sites don''t even run?Yes, it is very disappointing to see those entries not running and it doesn''t look good. At the same time, keep in mind the scope of the contest entry. 24 hours of development. Adding, "can it take a slashdotting" to the judging criteria is a bit of a tall order. Something bad can happen in any scenario and worrying about every what-if assumption leads directly to analysis paralysis. The fact that they''re not holding up to demand at the moment is still disappointing, but do not assume that application was designed to withstand the number of users that are currently hitting it.> If Railsday is being run on a dedicated server I can''t imagine why it would > be incapable of handling the load. I don''t think that it is getting > anywhere near a slashdot-sized crowd.>From using the site, it feels like the database is somewhat peggedright now and it is not surprising. The database is supporting multiple applications, each with their own requirements. How to support winning apps or all entries in the future is an important consideration that will/should get attention.> Another annoyance is that even if we try to download the site and run it > ourselves, just to see what it is like, we are extremely hampered, because > it is only available in an online svn browser. Please at least put up > tarballs of each site for us to download. >I kind of like that it''s currently available via subversion One, you can get a history of revisions. :) You can get a feel for the life of the contest entry and see how it evolved over the short time of development. I also like it because many people are new to source control, as others are new to MVC and new to Ruby and new to unit/functional testing and so on.This is just another one of those little things that I think the whole Rails "package" helps make people develop a little better. I look forward to another Rails Day despite whatever quirks occurred along the way. It was bumpy and ugly at points, but it was still a success in my book. Just my $0.02. John
> Now, I know that Rails can actually scale, and I very much appreciate all > those who are donating their time, energy and resources to make Railsday > possible, but has anyone considered the negative publicity that > _will_ occur > if the first and third place sites don''t even run? > > If Railsday is being run on a dedicated server I can''t imagine > why it would > be incapable of handling the load. I don''t think that it is getting > anywhere near a slashdot-sized crowd.Well before we start saying Rails can''t scale, I think we should see if we can get a hold of the server logs from the initial days when the winners were announced. Maybe we will see a rather large amount of traffic, that most scripting languages would not have been able to handle either. Conversley, maybe we will see that there wasn''t that much traffic, and that the sites being down is really an indicator of Rails'' scalability. The word on the street seems to be that Rails is great from a development perspective, but has serious scalability/performance issues. From what I''ve seen, these accusations aren''t completely unfounded. I think progress is being made. It will be interesting to see how much performance improves when the work that Stefan Kaes has been doing is merged in. Steve
Lucas Carlson
2005-Jul-04 06:30 UTC
Re: Railsday -- Demonstrating that Rails "can''t scale"?
The reason for the instability of the Rails Day winners was a confusion as to which version of Rails they were built to run on. Many of the sites would only run on beta gems and many would only run on the stable release. Switching the sites to the stable gem, they have not thrown an Application Error in days. I am sorry for the confusion, it took a little while to track down the root of this issue, but beta gems is admittedly unstable, it is beta after all. -Lucas Carlson http://tech.rufy.com/
Joe Van Dyk
2005-Jul-05 01:00 UTC
Re: Railsday -- Demonstrating that Rails "can''t scale"?
I thought basecamp was running off of beta gems? On 7/3/05, Lucas Carlson <rails-1eRuzFDw/cg@public.gmane.org> wrote:> The reason for the instability of the Rails Day winners was a > confusion as to which version of Rails they were built to run on. > Many of the sites would only run on beta gems and many would only run > on the stable release. Switching the sites to the stable gem, they > have not thrown an Application Error in days. I am sorry for the > confusion, it took a little while to track down the root of this > issue, but beta gems is admittedly unstable, it is beta after all. > > -Lucas Carlson > http://tech.rufy.com/ > _______________________________________________ > Rails mailing list > Rails-1W37MKcQCpIf0INCOvqR/iCwEArCW2h5@public.gmane.org > http://lists.rubyonrails.org/mailman/listinfo/rails >
David Heinemeier Hansson
2005-Jul-05 09:42 UTC
Re: Railsday -- Demonstrating that Rails "can''t scale"?
> I thought basecamp was running off of beta gems?Most of the core group is running on SVN/beta gems (and lots of other people too). It sounds unlikely that just that would cause anything to crash. And if it did, there would be a trace in fastcgi.crash.log. It''s even more unlikely since the errors where on/off. That said, apparently forcing all apps to use the same gem set worked. -- David Heinemeier Hansson http://www.loudthinking.com -- Broadcasting Brain http://www.basecamphq.com -- Online project management http://www.backpackit.com -- Personal information manager http://www.rubyonrails.com -- Web-application framework