atom feed41 messages in net.java.dev.hudson.devRe: Migration from java.net
FromSent OnAttachments
Kohsuke KawaguchiOct 2, 2009 2:03 pm 
Stephen ConnollyOct 2, 2009 2:22 pm 
Andrew BayerOct 2, 2009 2:33 pm 
Kohsuke KawaguchiOct 2, 2009 2:41 pm 
Dean YuOct 2, 2009 2:50 pm 
Kohsuke KawaguchiOct 2, 2009 3:02 pm 
Alan HarderOct 2, 2009 3:09 pm 
Stephen ConnollyOct 2, 2009 3:16 pm 
Frederic JeanOct 2, 2009 3:26 pm 
Kohsuke KawaguchiOct 2, 2009 3:36 pm 
Kohsuke KawaguchiOct 2, 2009 3:45 pm 
Andrew BayerOct 2, 2009 3:57 pm 
R. Tyler BallanceOct 2, 2009 3:59 pm 
Jason ChaffeeOct 2, 2009 4:18 pm 
Kohsuke KawaguchiOct 2, 2009 4:30 pm 
Jorg HeymansOct 3, 2009 5:25 am 
JotiOct 3, 2009 6:40 am 
JotiOct 3, 2009 6:41 am 
Jorg HeymansOct 4, 2009 12:58 pm 
Benson MarguliesOct 4, 2009 3:25 pm 
Stephen ConnollyOct 4, 2009 11:48 pm 
Olivier LamyOct 5, 2009 12:17 am 
Jorg HeymansOct 5, 2009 12:31 am 
Jorg HeymansOct 5, 2009 12:37 am 
Benson MarguliesOct 5, 2009 3:53 am 
Stephen ConnollyOct 5, 2009 7:43 am 
Michael DonohueOct 5, 2009 8:05 am 
Edelson, JustinOct 5, 2009 8:40 am 
Jason van ZylOct 5, 2009 8:44 am 
Kohsuke KawaguchiOct 5, 2009 9:48 am 
Kohsuke KawaguchiOct 5, 2009 9:54 am 
Andrew BayerOct 5, 2009 1:06 pm 
Benson MarguliesOct 5, 2009 1:53 pm 
Stephen ConnollyOct 5, 2009 2:46 pm 
Jorg HeymansOct 6, 2009 10:26 am 
Andrew BayerOct 6, 2009 10:36 am 
Julien RenautOct 6, 2009 10:50 am 
Jorg HeymansOct 6, 2009 1:26 pm 
Jorg HeymansOct 6, 2009 1:30 pm 
Edelson, JustinOct 6, 2009 1:34 pm 
Kohsuke KawaguchiOct 6, 2009 2:55 pm 
Subject:Re: Migration from java.net
From:Benson Margulies (bima@gmail.com)
Date:Oct 5, 2009 1:53:13 pm
List:net.java.dev.hudson.dev

While I'd hate to have Jason's sentiments or my endorsement of them get copied verbatim to the entire ASF membership, I think that I agree with him.

ASF's IP approach is really important for code that gets embedded in products, where you really want to know that someone isn't going to rise up like a submarine and start taking IP ownership potshots at you.

For a tool like Hudson, my opinion is that this is a much smaller problem. So using a lighter-weight approach to hosting the project makes sense to me.

I've found java.net to be unacceptable slow ever since I met it, so I find it hard to be an optimist. As for Kenai, I worked on a project on it, and I found that there were a lot of things that could only be done/fixed by the Sun people behind the curtain, in addition to the maven clumsyness.

Having said all of this ... if this community wants to investigate whether a relatively painless transition to ASF is possible, I'm willing to be the sacrificial negotiator. We clearly already have an active and diverse community. The big pain is the IP clearance issue of getting all the code licensed to ASF. And that may be more pain that his group wants to suffer.

--benson

On Mon, Oct 5, 2009 at 4:06 PM, Andrew Bayer <andr@gmail.com> wrote:

I am a fan of getting our own hosting - it would give us the advantages of hosting our own JIRA and Confluence servers that we already have, plus full control over all the rest of the hosting (including the distribution site, which is the main concern for moving away from java.net in the first place). That also plays into my longer-term dreams of having our own build/test infrastructure - if/when we could ever afford it.

I'd be willing to pledge, say, $500 up front towards hosting costs, with more down the road if needed.

My biggest concerns with Kenai are (a) the lack of a Maven repo - I know we could jury-rig one using SVN or WebDAV, but that seems far from preferable - and (b) performance. How do we know Kenai (or any other forge-type hosting situation) will scale well down the road? It'd be a shame to end up stranded on another over-utilized/under-supported service - Kenai may not be the right place for a project with as heavy usage as Hudson. I think I'd want a sense of where Kenai's planning to go, what kind of support we can expect, etc.

A.

On Mon, Oct 5, 2009 at 8:44 AM, Jason van Zyl <jas@maven.org> wrote:

Do not take the project to Apache, that's just a bad idea.

I believe as Jorg points out it is a powerful brand but not much more beyond that to be honest. I have been at Apache 8-9 years and if I were starting a new project I would never take it to Apache now. The incubator would just be a nightmare, and Apache's infrastructure is honestly not that great and neither is the legal infrastructure. I think you would be shocked at how poor both of those aspects are. It has none of the legal or project management rigour that Eclipse has. Not even remotely close.

Hudson as a name already stands on its own and isn't going to benefit from any branding power Apache has, and that's really all it has to offer IMHO. There are far too many people who will have opinions about how you should run your project while not doing much of the actual work. The overhead of dealing with the organization is too high. Your email to coding ratio will skyrocket. Keep your autonomy, just get a rock solid setup.

I know all the people at Contegix and I think they are awesome. For the cost of a machine and US300/month you get the best hosting in the world. No it's not a cheap-o Dreamhost setup, but you'll get 24/7 support, people responding in 2 minutes, they are experts with JIRA and Confluence, they host Codehaus, Maven Central, OpenSymphony and they are super OSS friendly. If you talk to Matthew Porter he might cut you a deal to get started. I'm sure between the core committers here and a little bit of fund raising we could get a machine and 2 years of hosting banked.

I'll ping Matthew Porter if you like and see what he might be able to to for you guys. Then the machine is yours, you can do with it what you like if you want to move it later. The important things like the maintenance of the machine and JIRA/Confluence upgrades will be taken care of. They are experts with qmail so you'll get all your lists as well. And you keep your autonomy.

If you wanted to pursue and use Contegix I would start with a donation of a machine and 1k toward future hosting costs. I have no affiliation with Contegix other then I use them and love them. You can find tons of references yourself but I can pass more along. Patrick Lightbody (Selenium) also uses Contegix for all the Selenium project related hosting.

On 2009-10-04, at 12:58 PM, Jorg Heymans wrote:

On Sat, Oct 3, 2009 at 3:40 PM, Joti <joti@gmail.com> wrote:

to give my two cents: -1 for Google Code.

They dont have JIRA, they're not Sun-connected (so Kuhsuke has less possibilities of doing something if anything goes bad or something custom/special is needed) and to me Kenai looks quite ok. And since Hudson has official Sun support, maybe the project will get some more attention from the Kenai makers.

Well I don't follow your passion about Jira, but i guess that is a matter of taste :) I also don't follow your argument about needing a Sun-connected hosting partner. A lot of OSS projects are hosted independent of the company of its creator and most of them are doing just fine without having a guy on the inside that can fix things. As a project you should not rely on one guy only being able to do custom/special things in case something goes wrong, what if Kohsuke is no longer interested/able to work on Hudson at some point ? Are you hoping that the next core hudson contributor is another Sun employee that can pull strings from the inside ?

Just a crazy thought: why not apply for incubating Hudson at Apache, one of the strongest, most independent oss brands of the moment ? That would secure its OSS future for a long time, and since they're using Hudson already (http://ci.apache.org/) it might even attract a host of new developer talent. [[full disclosure: i was a commiter on the Apache Cocoon project a few years ago, so i might be a tiny bit biased myself here]]

Thoughts ? Or has the Kenai project been created already ? :-P

Cheers, Jorg

Anyhow: Thanks for Hudson and the support that is already there from all of you :D

Cheers, Joti

...who is hoping on something better (JIRA) than Bug/Colabzilla for a long time to have more motivation to actually file the bugs he found.

2009/10/3 Jorg Heymans <jorg@gmail.com>

On Fri, Oct 2, 2009 at 11:03 PM, Kohsuke Kawaguchi <Kohs@sun.com> wrote:

As we know, lately java.net has been suffering constant performance problems and minor outages. Because of this, the topic of moving the project from java.net has been a recurring topic on this mailing list, dating back at least before JavaOne 2009.

And this week, java.net had the biggest outage ever. It rendered the entire site inaccessible for about 3 days, and the site is still inaccessible to many users in Europe, Latin America, Russia, and Asia, for which we still have no ETA as of this writing.

During this period, plugin downloads and updates were unavailable, new users weren't able to try out Hudson, and committers weren't able to push changes. Furthermore, because of a problematic code in Hudson, this outage rendered Hudson installations in the wild unconfigurable [1].

What I'm hearing from the Hudson community is that this is just not tolerable. So I think it's time to move off from java.net --- This is putting me into a very hard position, but I think I have no other choice, really.

The hard part is to figure out the migration plan. This is where I invite the community for suggestions and discussions.

Kohsuke, thanks a billion for finally sharing your thoughts on this. As i've said a few times in the past, i cannot agree more on migrating away from java.net - let's get Hudson flying again !

As to the currently suggested alternatives I must say that Kenai looks good at first sight but i'm wondering if Hudson really needs to be the guinea pig for a new OSS hosting offering ? It seems to still be very much 'work in progress' (correct me if i'm wrong) and i don't know if Hudson should take another potentially bumpy road after the java.net one. Also, do we know what will happen to Kenai once the Oracle deal is completed ? Perhaps something to take into consideration.

Has anybody given thought to code.google.com as base for hosting ? Piggybacking the google infrastructure services behemoth is bound to keep us up to date with some of the latest and greatest in development tools and collaboration. It might not have all the features upfront in the way that we require but it wouldn't hurt to investigate i think. I'm willing to put in some time to check out the infrastructure and features and try to match them to what we require, unless ofcourse there are severe objections upfront against hosting anything at google ?

Other than that, i'm ofcourse offering any evening/weekend time required to help out in the move, regardless what the decision is.

Cheers, Jorg

Thanks,

Thanks,