Why OpenSolaris Failed To Build a Community

xtaski writes "Ted Ts'o, one of the earliest Linux developers, points out some serious flaws in OpenSolaris. There is a severe lack of developers, for one. Apparently, after 3 years, the OpenSolaris 'developer community' is still struggling to get the proper tools for developers to develop! Ted also points out some other flaws which make it clear just how disconnected the executives at Sun are from what's really going on in their 'open source communities.' He notes, 'It was never ... Sun's intention to try to promote a kernel engineering community, or at least, it was certainly not a high priority for them to do so.'"

Comments

For those too lazy too read the article:

[SatanicPuppy]

The answer is: "They acted like a bunch of dicks."

OSS is a labor of love. You've got to want to work on the project, and you've got to be able to work on the project.

If you put a big chunk of your time into something and get rudely dismissed, then its hardly likely that you'll continue to contribute.

Re:For those too lazy too read the article:

[afabbro]

Where I work we have two sun mainframes;

Um, no you don't.

Re:For those too lazy too read the article:

[ryanov]

I would probably choose Solaris over Linux on Sun hardware. FMA, ZFS and ldoms (depending on the machine of course) are pretty cool.

Re:For those too lazy too read the article:

[Kent Recal]

Hm, I'll go through them one by one.

Zones are indeed amazing but we don't have a use-case for them on our
production hosts. Maybe we could have used "zone-images" to deploy our
stuff but that'd really only be an excuse to use zones for a task
that seems to be better suited to apt (dependency tracking, controlled
distribution of updates etc.).

ZFS, no doubt, is something we miss.

SMF. Well, as often with solaris I love the concept but hate the implementation.
Yes, it beats sysv-init hands down. But, gah, XML, and many of the pkgs on solaris
freeware don't even support it yet. We learned how to roll our own pkgs (with SMF etc.)
but the lack of a package-manager akin to apt turned this into a time sink.

DTrace is a similar story. Ofcourse it is powerful but on a every day base we simply
don't need that power. We prefer the simplicity of strace which has, so far, been
good enough when we needed it and never forced us to wade through documentation
for supposedly "simple" tasks like dtrace did. Dtrace alone seems to be basically
useless until you wrap it up in a fairly elaborate script for the task at hand.
Yes, "DTraceToolkit" provides these scripts for many common tasks, but not for all.
Looking back at the first dtrace tutorial that I found (http://developers.sun.com/solaris/articles/dtrace_example.html?feed=DSC)
I can still only shake my head about the effort it takes for such a simple task.

About messing on the command line often: No, we don't need that "often" but regularly.
We install stuff, upgrade stuff, like everyone does. The major gripe was: We don't want
to repeat ourselves for n hosts. Ideally we want to roll packages and distribute them
in a controlled manner. Solaris left us out in the cold here and implementing a 3rd
party framework like puppet feels wrong when linux never even made you think about that.

No, pkgadd -d is not too hard for us. But manually downloading each dependency from solarisfreeware
or whatever source and installing it on each individual host is. Not too hard as in "we can't do it"
but too hard as in "dude, that feels backwards".
Feel free to enlighten me. How do *you* upgrade curl (random example) on 8 solaris hosts?
In debian we review the distro pkg and roll our own when necessary, test it, push it to the
mirror and the nightly update pulls it to all hosts. What tools did we miss that enable
an equivalent workflow for solaris?

Generally I wouldn't say we weren't able to handle solaris. I'd say it turned out we didn't *want* to.
It's very well possible that AIX would break us into a quivering goop. For that precise reason I have
no intention to ever touch it? It has nothing to offer that interests us, so why should I even
consider it?

Well and finally, your last statement makes you sound strikingly similar to the Sun sales reps
that we had. They also kept trumping the horn of "use our modern OS with features not
yet available for linux" to "utilize our kickass hardware to the fullest".

The problem probably lies in the definition of "utilizing our hardware to the fullest".
For you and Sun that probably means to squeeze out the last 10% of performance.
For us it means being able to fluidly run and maintain the hosts (that includes
a sane update-procedure etc.) with a minimum cost in terms of manhours.

In other words: The time we saved on "solaris administration overhead" since
we abandoned it quite likely paid our last 3 xfires alone.

You may blame it on our lack of solaris expirience or even on your perceived
"ineptitude". Still it was the right decision for us.

OpenSolaris fails to build community b/c it sucks

[Jeffrey Baker]

You have to have a good product before you can have a community. Linux built its early community based on tinkerers and hackers who found it easy to play with. Early Linux distributions, you may recall, were all inclined to integrate well with DOS. Some of them could even be installed _in_ DOS. You could install Slackware and be up and running with an editor and compiler in half an hour. OpenSolaris doesn't follow this example. Using it is a tremendous pain in the ass. Its installer runs for 2-4 hours on the midrange PCs I've tried to install it upon. Once it's "installed" you still have to grope around trying to find familiar tools, which are maybe under UCB or perhaps under GNU subdirectories. It's hard to download software from the 'net and ./configure it. Hardware support is very thin.

To get a hacker community, you have to offer fun. OpenSolaris is simply not fun. It reminds me of work.

Instant success

[mooreti1]

Yes, because Linux was such an instant success! Wait...no, it wasn't. Everyone forgets that any community, either real or virtual, takes time to build. I believe that counting OpenSolaris as a failed community is premature, at the least.

Re:Instant success

[nrozema]

I guess success vs. failure comes down to what your initial expectations were.

If you were looking for Yet Another Open Source Linux Replacement, and have failed to receive it from a barely four year old project, then sure, I suppose to you that project has "failed".

If, like me, you saw OpenSolaris as a sandbox and open dialogue with the community to shape the next version of Solaris, and not a Linux replacement, then perhaps you aren't so disappointed at the moment.

The development of ZFS in particular has come a long way in later builds of OpenSolaris vs. what you'll find in Solaris 10. The previous development model would have seen that happen in a vacuum, without community interaction and contribution. That alone is a success in my mind.

Re:Instant success

[Anonymous Coward]

Oh please, you're comments are so wrong and out of date.

This whole Sun is faking open source is just BS.

From the very beginning, Sun had been a very open company. Look at one of the original founders of Sun, Bill Joy.

You're obviously someone that came late to the game and are just talking out of your ass.

This is my opinion, but I believe it to be based on a bit of fact. Without Sun, from the early days, pushing for open standards, and their publishing of many of these standards, I don't know how Linux could have come to the point where it could be today.

What it always seemed to me was Sun was interested in doing with Unix was to define a common platform that could benefit all Unix. As the tide of Unix rose, all could benefit. Sun's position was something like, we can build a better boat.

Though there have been cases where certain people at Sun have been hostile to FOSS there are just as many if not more that were embracing of it. I don't know how McNealy got this "cound't stomach" FOSS image. Maybe calling linux a "toy os". But really, if you compared Linux to Solaris at that time, he wasn't far off. Sun has open sourced a lot of code and helped a lot of open source projects. The only people that seem to not like Sun are some linux zealots.

Re:Hmmm - Linux Fan Boys speak out....

[keithjr]

As it turns out, Linux development community members are critical of competing operating systems. How is this news?

The point of picking and choosing your operating system is so that you can pick the best tool for the job. If that tool is Solaris, then use it. If not, so be it.

Since Sun actively develops Solaris (and thus parts of OpenSolaris), do they really need individual contributors?

Re:Not welcoming your Scott McNealy overlord?

[SatanicPuppy]

Actually, no.

The lesson here is: If you're going to try to court people active in OSS development, then you're going to have to be nice to them, and you're going to have to let them take some ownership.

IBM is being smart; they're reaping rewards far in excess of their investment. Effectively they've outsourced their development, and while the terms of the "outsourcing" say that they have to share everything that comes out of the project, they're still in a position to steer, and support the product.

I'm not sure how you equate that with "control"; sounds just like more FUD to me.

of course not

[nguy]

OpenSolaris was an attempt by Sun to throw some sand in the gears of Linux, not to build an open source project. They are doing the same thing with OpenJava.

I mean, who is going to contribute to such a project if (1) Sun engineers keep calling the shots, and (2) anything you contribute needs to be given to Sun so that they can sell it to paying customers?

If Sun were serious about making Solaris and Java open source projects, they'd release them under a single, open source license only. That would probably have to be BSD.

And why not? Solaris was BSD licensed to begin with; it was Sun that made it proprietary.

Reasons for this failure.

[suck_burners_rice]

The disconnection between Sun's executives and the kernel developers might be one reason why OpenSolaris is failing to build a community, but I believe a much larger reason is the lack of any substantial need for OpenSolaris in the market at this point. Currently there is so much development around Linux and the BSDs that these projects fulfill most of the users' needs and offer people looking for an OS something quite compelling, with a developer community in the millions of knowledgeable people. OpenSolaris is first and foremost suffering the chicken-and-egg problem that since there isn't much of a developer community, nobody wants to join, and secondly, since Linux and the BSDs can carry out nearly all the functions that OpenSolaris carries out, there's no compelling need for developers to join that community. Let's face it, Sun should concentrate their efforts on improving Linux and selling distributions and support for their custom distribution. Part of this improvement would entail porting the few advantageous features that Solaris has over Linux currently. OpenSolaris would eventually be phased out completely. Otherwise, they are simply throwing good money after bad.

Download barriers

[SgtChaireBourne]

Downloading is a royal PITA. The registration is usually a deal-breaker. Almost nothing I've ever run across that's worth anything requires registration for download. However, as a (former) long-time Solaris / SunOS user and major FOSS user, I felt compelled several times to try to circumvent that. But then there's no real way do a network install and othewise week download choice.

That gripe aside, the article is a bit premature. Though time is running out and it could become true if Sun decides to keep downloads off of anonymous FTP, AFS and Bittorrent.

Because Solaris without Sparcs isn't that great?

[Kenja]

Only reason I ever used Solaris was for the Sparc hardware. Soon as Sun went Intel based, they where dead to me. Why spend more money for the same level of hardware when the OS has less support then Free(tm) options?

As a member of the community...

[saleenS281]

I guess I disagree. I'm on several of the opensolaris mailing lists, and they're ALWAYS busy. And not just with people from Sun, people from all walks of life. To claim that opensolaris has failed is preposterous to me. I guess I don't quite understand what this mans idea of *success* is, but apparently having users and contributers from both sun and the public abroad isn't *success*.

Is his complaint that the majority of code comes directly from Sun? If so... let me just say *DUH*. If you have thousands of PAID programmers writing code, nobody is going to waste their free time re-writing from scratch. On the flip side, there's TONS of public side-projects, I can think of several around zfs like the automatic snapshots. Or maybe that little side project called nexenta.

I think I understand what his issue is... he doesn't even know what the opensolaris community is. By his definition, one distribution of linux is a measure if its success or failure. Last I checked, when we talk about linux, we're encompassing ubuntu, redhat, suse, slackware, etc, etc, etc... Guess what, the same holds true for Opensolaris.

So... basically, it sounds like a linux zealot casting a stone because he's most likely upset that Sun wont' release solaris under the GPL so that linux devs can start ripping code.

Re:As a member of the community...

[tytso]

Please let me make this clear. I was not disparaging Open Solaris as an operating system. And I was quoting Jon Plocher, a Sun Engineer working on Open Solaris, when he admitted that Sun didn't get the community they were hoping for. So it you can call it failure in terms of Sun being to get the results that it had hoped for when it released (most of) Solaris under an Open Source license. Other people who were major Solaris fans, and who were excited with whatever scraps Sun might throw from the table, might be mightly pleased with what Sun did. But nevertheless, it is interesting that Sun hasn't achieved what they hoped to accomplish with Open Solaris after three years.

The reason why I found Jon Plocher's candidate statement for the Open Solaris Governing Board so interesting was that it was first that I had seen someone from inside Sun comment about the what Sun had been hoping to achieve by release Solaris under a Open Source license that didn't appear influenced by Sun's marketing/spin machine. I don't believe Sun's officially stated reasons (that show up on the CEO's blog, for example) because after three years their words have not been matched by their deeds.

So for me, it's more about correcting the marketing spin. If Sun salescritters want to pay analysts to create Total Cost of Ownership white papers which compare the cost of the most expensive get-someone-on-the-phone 24x7 Red Hat support with a support-by-email Solaris support subscription, I might mock their desperation.

Similarly, if Jonathan Schwartz wants to talk about how wonderful it will be that Open Solaris is Open Source, and how they will reap the benefits of having Open Source developers, but three years later still have processes that result in 0.6 patches/day being accepted into Open Solaris, then I think it's only fair that to point out the chasm between his words and his company's actions.

OpenSolaris is a very young opensource initiative

[blind biker]

Give it some time - it's still growing, and while there are some adjustments to be made, the situation is far from catastrophic for its stage of development. After all, there's a number of people contributing to it, and hopefully as processes and community contacts improve, the contributors will increase in number. You have to take into consideration that it's a huge chunk of code and some people are still just lurking to find their place under the sun (no pun intended).

OpenSolaris is an interesting operating system, I don't doubt it'll grow in popularity among developers, however slowly. As I said, give it some time, we have only just begun.

Linux Partisan Disparages Non-Linux OS

[planetralph]

This is news?

Re:OpenSolaris fails to build community b/c it suc

[thsths]

> So where does OpenSolaris fit in? It seems to be an OS lacking a niche.

The niche for OpenSolaris is the 64way mission-critical server. Unfortunately, even ultimate kernel hacking enthusiasts rarely have one of those at home.

Re:I though it was just because it sucks

[rumith]

There is some truth in GP's words. While Solaris is full of great ideas, and most of them are pretty well implemented, one must admit that the native userland hardly differs from what it has been in early '90s. Of course, the GNU environment is available, but it hasn't replaced the default one so far. Once that happens, I expect Solaris user base to boost.

Re:mirror

[Anonymous Coward]

I was really following along with you until you did the "kiss a girl" quote. I followed the link and noticed the email is from 1996. That sort of destroyed your credibility about the OpenSolaris development right there.

Re:Hmmm - Linux Fan Boys speak out....

[catmistake]

I have to agree with parent and grandparent posts... Seems to me the community is thriving. I'm sure I'm missing the point of OSS somewhere, but with OpenSolaris I see it as an attempt or ambition to get the rock-solid stability of Solaris and Sun hardware for free on most any hardware, and the ability to initiate experience with Solaris on the cheap (however, Solaris itself is a free download, just licensed propietary, so why practice street ball when you can play stadium? OK poor analogy).

Re:OpenSolaris fails to build community b/c it suc

[Anonymous Coward]

> nobody really knows what NetBSD is about

Running on everything and having really elegant source code.

> So where does OpenSolaris fit in?

A stable ABI, with extensible kernel constructs that don't require rejiggering the ABI every couple weeks. You write your solaris driver once. Ever.

CDDL

[Anonymous Coward]

If Sun had chosen GPL, then direct code sharing with Linux.

If Sun had chosed a BSD license, then there could be direct exchange of code with a BSD OS.

Picking CDDL was design to keep the control, but I did not help spread its use.

Re:CDDL

[BrainInAJar]

Picking CDDL was to keep people from taking the code like they did with the original 4.xBSD series, while still allowing people to write closed-source drivers for it when it gets popular enough to attract people to do that.

Imagine how good the hardware support could be on Linux if Linus had just picked a license that lets people link in to the kernel with non-GPL code, and kept a stable ABI so that it made sense to do so. THAT'S what Sun was after

It's Difficult to Download OpenSolaris (rants)

[turgid]

I spent over a week trying to get Open Solaris build 85. Sun just doesn't get the free distribution thing. You have to register and log in to the Sun Download Centre, from where you can download the CD or DVD images. They try to persuade you to get the Sun Download Manager which is some Java app that gives you pause and resume buttons for the download.

I tried 5 or 6 times to download on different days with the download stalling at sometimes as much as 90%. On the 8th day, I got the whole image. So much for their download manager. You just have to overwrite the chunk you have and start again.

After all these years, they still haven't sorted out the auto-layout of the filesystems. There's not enough room partitioned to install their developer tools.

I went to build gcc-4.2.3. That took 5 days and about a day of CPU time. OK it's an ancient 500MHz USIIi that I got for nothing, but...

See, Solaris's /bin/sh is badly broken (archaic) and can't be used to build gcc. So you set CONFIG_SHELL to be ksh. Only the configure scripts in gcc are still broken from gcc-3.1.x days and two of the scripts it generates, bin/as and bin/collect-ld at each stage of the bootstrap are broken because they begin #!ksh instead of #!/usr/bin/ksh or whatever.

When I used to build gcc on Solaris, I just sed'd all the scripts to replace /bin/sh with /bin/bash or whatever.

So, for the casual SPARC/Solaris power-user/Linux developer myself, it's just too darn inconvenient.

And the stuff in /usr/sfw/bin, which is where the "Open Source migration" into Solaris proper was supposed to happen still looks like it did in 2005, 3 years ago.

Solaris has a brilliant kernel. Putting the DVD images on Bittorrent (officially) like OpenOffice.org, would be a great start. There are too many hurdles for the average user to go through who might have been interested in trying it out. I don't have to register to download Slackware, Ubuntu, KNOPPIX, NetBSD etc.

Sort out the default install so that the disk layout is sane and make it trivial to install the GNU toolchain.

But I've been through all this years ago, and it pains me to see that it still hasn't been fixed.

Typical Sun

[codemachine]

Their OSS stuff is generally high quality production code: Java, OO.org, Solaris...

But the OSS projects themselves have problems involving people outside of Sun. In the case of OpenOffice, Novel had to fork to get their improvements in at a reasonable pace. NeoOffice had to fork to get a useable Mac OS X version at all.

OpenSolaris may head down the same path, with Nexenta having the better and more available build than the main project does.

It seems that Sun knows how to code. They just don't know how to be open. Websites with registrations and download managers are barriers. Projects that accept outside contributions at a glacial pace, if they accept them at all, are barriers. And these are typical of everything Sun.

If they could learn how to create vibrant open communities, while still retaining the ability to guide/control the projects as much as needed for their purposes, they'd be an even more incredible force in the OSS world.

Re:I though it was just because it sucks

[Anonymous Coward]

> There are suspicious java processes running all the time, too.

Solution:

svcadm disable webconsole
cacaoadm disable
cacaoadm stop

Re:OpenSolaris fails to build community b/c it suc

[Secrity]

Somebody with a mission critical server should be running real Solaris, not OpenSolaris.