This is kinda depressing. Not so much that Alasdair quit as his reasons.
http://lwn.net/Articles/514046/
http://lwn.net/Articles/514046/
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
Openindiana lead resigns
- Thread starter danswartz
- Start date
coolrunnings
Limp Gawd
- Joined
- Nov 21, 2004
- Messages
- 511
That is depressing. 
(Disclaimer: I am a user of Solaris 11, Linux and a lot of open source software. I like FOSS a lot)
Sounds to me like he quit because he does not have the time to do the work anymore. He seized the opportunity to complain that the world does not adhere to his ideal open-source worldview.
These few things are nagging at me:
1)
Nexenta and other users of the OI code are companies. They are in business to make money. How is contributing to an open source project (and thus giving away their development time to their competitors) going to make them money? Expecting that this will happen is naive.
2)
He complains about the implementation of ZFS and other features in Linux. It's like he does not want Linux to get better because it would make his project more obscure. This seems just spiteful to me. On the one hand he does not want features in Linux. On the other hand he couldn't even get it to boot without help from the Linux kernel team:
"Lastly, despite their lack of a handle on what's happening with Unix/Linux distros in the real world beyond kernels, I'd like to thank all those who have contributed to Illumos, without which OpenIndiana
would not boot."
Even in his thank you note, he couldn't help but point out that they don't have a handle on what happens in the real world...
3)
"Add to that our increasingly out of date and poor hardware support due to the march of never ending new LAN/SATA/SAS/motherboard/GPU chipsets and you start to get the picture."
Maybe if they were not targetting the desktop, the variety of hardware that they had to deal with would not be so large? It looks to me like the majority of server grade hardware (at least the stuff that matters) has solaris drivers made by the hardware vendors. Oracle Solaris seems to be doing fine with regards to hardware support.
4) (closely related to 3)
A company the size of Canonical is trying to create a desktop OS equivalent of Windows. Despite their good efforts, Ubuntu is not even close to achieving that goal (I wish it wasn't so). If a company with Canonical's resources cannot achieve this goal on a much more mainstream platform, then why does the OI project think they will be able to with just a small community of developers that work for free.
It's sad that he quit. I wish OI and Solaris in general were more widely used. That does not mean I am as blind to reality as Alasdair seems to be.
Sounds to me like he quit because he does not have the time to do the work anymore. He seized the opportunity to complain that the world does not adhere to his ideal open-source worldview.
These few things are nagging at me:
1)
Nexenta and other users of the OI code are companies. They are in business to make money. How is contributing to an open source project (and thus giving away their development time to their competitors) going to make them money? Expecting that this will happen is naive.
2)
He complains about the implementation of ZFS and other features in Linux. It's like he does not want Linux to get better because it would make his project more obscure. This seems just spiteful to me. On the one hand he does not want features in Linux. On the other hand he couldn't even get it to boot without help from the Linux kernel team:
"Lastly, despite their lack of a handle on what's happening with Unix/Linux distros in the real world beyond kernels, I'd like to thank all those who have contributed to Illumos, without which OpenIndiana
would not boot."
Even in his thank you note, he couldn't help but point out that they don't have a handle on what happens in the real world...
3)
"Add to that our increasingly out of date and poor hardware support due to the march of never ending new LAN/SATA/SAS/motherboard/GPU chipsets and you start to get the picture."
Maybe if they were not targetting the desktop, the variety of hardware that they had to deal with would not be so large? It looks to me like the majority of server grade hardware (at least the stuff that matters) has solaris drivers made by the hardware vendors. Oracle Solaris seems to be doing fine with regards to hardware support.
4) (closely related to 3)
A company the size of Canonical is trying to create a desktop OS equivalent of Windows. Despite their good efforts, Ubuntu is not even close to achieving that goal (I wish it wasn't so). If a company with Canonical's resources cannot achieve this goal on a much more mainstream platform, then why does the OI project think they will be able to with just a small community of developers that work for free.
It's sad that he quit. I wish OI and Solaris in general were more widely used. That does not mean I am as blind to reality as Alasdair seems to be.
Does this change anyone's opinion / decision on using OI / Napp-it for a home built, virtualized NAS? I wanted to go with FreeNAS 8 but ran into limitations within an ESXi All-in-one. I was just about to go ahead with OI / Napp-it but now am not sure what to expect is the better long run approach.
As with any other open source project where the lead developer(s) decide to leave, either new people will step up or the project will die a slow/fast death as users abandon that platform for something else. The project(s) that are adopted by the users may offer some/most/all the features of the original project or be the new shiny, but rudderless projects (with few exceptions) usually don't grow much once the leads and other contributors start to to jump ship.
brutalizer
[H]ard|Gawd
- Joined
- Oct 23, 2010
- Messages
- 1,602
There is Illumian, which seems to be the new distro, taking over OpenSolaris. And there is also Nexenta. Among a few others.
opensolaris as a desktop always was dead and always will be dead. the few people working on it were and are delusional.
as a niche server, the opensolaris code is doing fine, because just like with linux, there are companies sponsoring it. but as the guy above said, if ubuntu is going nowhere as a desktop OS, then solaris has ZERO hope.
as a niche server, the opensolaris code is doing fine, because just like with linux, there are companies sponsoring it. but as the guy above said, if ubuntu is going nowhere as a desktop OS, then solaris has ZERO hope.
I am not sure I agree that ubuntu is going nowhere, I have talked to many people here that have switched to ubuntu, and some even switched again to mint afterwards. They have loved the switch from windows.
These people have never used any linux before, and don't work with computers at all. So it has made headway here in the USA atleast, with the normal consumer.
These people have never used any linux before, and don't work with computers at all. So it has made headway here in the USA atleast, with the normal consumer.
I think the whole "Desktop market share" thing is a strawman.
Alasdir wasn't planning to leverage Time Slider and Gnome 2 into desktop dominance. He was trying to perpetuate an Open Source Solaris. And he was trying to give n00bz, mostly Linuxers, a comfortable way to ride the badass firebreathing beast that is Solaris.
And the fact that some of the people A. L. referred to in his LOR are sitting around generating strawmen and washing their hands of an Open Source Solaris (with a desktop or no) rather indicates he was not far off the mark in his last...
Alasdir wasn't planning to leverage Time Slider and Gnome 2 into desktop dominance. He was trying to perpetuate an Open Source Solaris. And he was trying to give n00bz, mostly Linuxers, a comfortable way to ride the badass firebreathing beast that is Solaris.
And the fact that some of the people A. L. referred to in his LOR are sitting around generating strawmen and washing their hands of an Open Source Solaris (with a desktop or no) rather indicates he was not far off the mark in his last...
Last edited:
Doesn't change my opinion, which is: as soon as something else is able to implement most of the ZFS features reliably (either linux with btfrs or windows with refs), then I'll switch away from solaris. I don't need/use zones or dtrace or any of the other well known features, I just love what ZFS can do. But there are just enough differences/quirks to be a pain sometimes with solaris that I'd rather use something else.
Many companies contribute to FOSS. Redhat, Canonical, and Intel, just to name a few, have put a lot of effort into open source software and make(d) money. And in "2)" he was talking about the Illumos team not the Linux kernel devs.
Personally, I have been testing ZFS on Linux and will be moving my storage to it. I am much more comfortable with Linux and it's much simpler to be familiar with a single OS.
"Doesn't change my opinion, which is: as soon as something else is able to implement most of the ZFS features reliably (either linux with btfrs or windows with refs), then I'll switch away from solaris. I don't need/use zones or dtrace or any of the other well known features, I just love what ZFS can do. But there are just enough differences/quirks to be a pain sometimes with solaris that I'd rather use something else."
Amen.
Amen.
Monkey God
Mangina Full of Sand
- Joined
- May 7, 2007
- Messages
- 6,723
After reading that whole rant, he sure is bitter.
Well, maybe in a few years they'll re-engineer ZFS to be compatible with the GPL... But they may have to port the 2 ZFS commands into Linux.
Monkey God
Mangina Full of Sand
- Joined
- May 7, 2007
- Messages
- 6,723
Well thats one of the many risks of being a major open source contributor/leader. You may not get help, and people may go elsewhere for their needs.
I setup a OI setup at home and experimented with it for a while...it worked pretty good.
Doesn't change my opinion either because I too went with Solaris for my home NAS. Similarly, it's not about the OS for me: I'm buying into ZFS, not Solaris v Linux v Illumos or whatever.
The real issue is that I really consider ZFS itself an Oracle owned standard and any fork outside of that I'm not likely to be interested in for a variety of reasons.
But OI does not need to die; maybe it just needs a better leader. However, as a manager in IT perpetually looking for talent, I find it hard enough to hire good people at market rates, let alone find the type of technical leadership skills that would be necessary for a project like this on a volunteer basis.
IMO, OI would be better off if a for-profit vendor truly owned it while retaining the open source philosophy that Sun originally had. Whether or not that's long-term viable is another issue, but it'd be an improvement over the current fragmented approach.
ZFS on linux is around us, and GPLed ....
major functions are already in zfsonlinux,
I plan to switch back to zfson linux from OI-zfs
... Hopefully.. ZFS would be in main kernel stream soon or later......
better waiting ZFSonlinux( many improvements as now) than btfs ( slow development as now)
Christopher
Gawd
- Joined
- Sep 16, 2002
- Messages
- 634
Also FreeBSD works great for everything I've thrown it at which has ZFS v28 in the current stable releases.
ZFS on linux is around us, and GPLed ....
major functions are already in zfsonlinux,
I plan to switch back to zfson linux from OI-zfs
Hopefully.. ZFS would be in main kernel stream soon or later......
better waiting ZFSonlinux( many improvements as now) than btfs ( slow development as now)
Since at present you evidently lack empirical data, please allow me to help fill you in on some of the subtexts to the press releases that apparently serve to inform your post.
zfsonlinux - a US DoD funded cash cow brought about by the GPL and that Lawrence Livermore Labs plans to milk to perpetuity - might be a tad BETA yet for the Linux mainstream.
Indeed, zfsonlinux may be a tad BETA for even Linux application developers, which may explain why porting ZFS to the GPL had to be socialized in the first place - big job, little pay.
Like said, "A few years".
BTRFS? A big improvement for Linux, but...
http://rudd-o.com/linux-and-free-software/ways-in-which-zfs-is-better-than-btrfs
Solaris 11 daily driver here.
Add to your comments that the fact that S11 has an easy install, a Gnome2 desktop, the best VBox-ability, better Zones, better networking, and a comfy front end to IPS, Oracle mostly accommodates the home tester/developer with the most comfy smooth introductory and/or test/dev Solaris experience on a desktop...
These Open Source Solaris techs are ahead of their time and solve problems Linux has just started to encounter (OpenStack anyone?). ZFS and clouds is $$ and everyone knows it. Hence certain people who used to be into Open Source who now feel that supporting an Open Solaris desktop distro is giving away the farm.
*(In accordance with the purposely vague license terminology)
/dev/null
[H]F Junkie
- Joined
- Mar 31, 2001
- Messages
- 15,182
++
This is what I use and has been around for YEARS (07 ? 08 ?). FreeBSD is also widely used.
I found FreeNAS and FreeBSD to both be buggy as hell. No where near as good as Solaris or OI.
This may be part of the reason BSD doesn't dominate the ZFS spaces.
If you're an application developer and the application isn't ZFS related I don't particularly see how ZFSonLinux is going to hinder or help their situation. I've used OI and it has it's fair share of bugs as well.
I used OI for I'd say 4 months or so to figure out whether I was going to use it at home or work. The biggest problem is that OI doesn't support many of the programs that even Solaris supports. I'm not talking about Gnome/desktop stuff either. I'm talking about things like backup solutions. OI is relatively limited compared to Solaris and it's no contest at all comparing either to Linux. What makes it worse is that the programs that are compiled for OI usually aren't as stable as other solutions which compounds the issue. The lead was absolutely correct.
For work I would go straight up Solaris. For home? I would do ZFSonLinux. The additional programs really add a lot if you know you're way around Linux.
/dev/null
[H]F Junkie
- Joined
- Mar 31, 2001
- Messages
- 15,182
DetailS?
FreeNAS is vastly different than FreeBSD. I've found FreeNAS buggy. I haven't found FreeBSD buggy.
Rsync and NFS and cron have kicked ass for years, if you know your way around Linux...
All this speculation that ZFS on Linux "just works". Hint: if ZFS worked on Linux without significant problems, Oracle would have already have it on Linux, or bought Red Hat outright. And I'd probably use it too.
I've spent a few months with OI151 on bare metal myself. It's buggy and unsupported, and the irc isn't much help to a new user. I use S11 for a lot of the reasons you and Alasdir mentioned.
Doesn't mean OI, as the only free and open source Solaris, isn't worthwhile.
I've found FreeBSD to be a genuine PITA to get anything running on ZFS.
Nevermind the day-long installs, the mysteries of package ports, the lack of virtualization capabilities, the onerous Zones implementation, and all the fun of dealing with semi-supported hardware - BSD in general seems to be best in purpose-built distros doing a speciality. Like pfSense or FreeNAS.
Unless you want to devote a few months or years, BSD's biggest plus is its license and its specialized distros.
And no, no BSD yet has as good an implementation of ZFS 128 as OI.
Alasdir is right: OI may suck but it doesn't suck as bad as the free alternatives.
Perhaps you should just end the speculation then and tell everyone what significant problems ZFS on Linux has!
Hmm I wonder if we can still change repositories and upgrade to Solaris 11 Express.
Rsync by itself isn't a backup strategy when you are dealing with large datasets. There's a reason why Solaris supports just as many backup solutions as Linux and the reasons aren't imaginary.
I don't believe I said that. But I am using for it home use and I've experienced no more issues than when I used OI. It's not speculation I'm using it now. Different issues? Yes. More? Nope. It's OK for home, but I wouldn't put it into enterprise use. At least not yet. We'll see.
Oh no OI is worthwhile. Just because support isn't where it should be doesn't mean OI doesn't have benefits. The problem is unless they fix the support issues I'd almost rather accept the limitations of SE then go the OI route.
Last edited:
I've run ZFSonLinux for well over a year now on a home raidz mythtv backend/media server. It has been super stable, except for 3 things....
1) At first, I only had 4 gig in the home server and with just moderate "desktop" use on the machine, the ZFS arc cache consumed too much memory, making swap thrashing. Fixed temporarily by manually setting a max arc cache size.... and I believe permanently fixed by a later release that by default not allowing more than 1/2 of total ram to be used as ARC cache.
2) For a long while, after every reboot I would need to execute a "zfs mount -a" to get the device mounted. Again, recently updated to new release and have no problem.
3) I use ZFS as a root filesystem at work, and sometimes after a ZFS/Kernel update, initramfs cannot mount the filesystem. To fix, I boot another zfs capable OS (another HD or livecd), do a force import, and then an export, then it works fine.
The snapshotting and the zfs send/receive by themselves are worth it. The ability to use linux apps/desktop/supported hardware makes ZFS on Linux just simply excellent.
1) At first, I only had 4 gig in the home server and with just moderate "desktop" use on the machine, the ZFS arc cache consumed too much memory, making swap thrashing. Fixed temporarily by manually setting a max arc cache size.... and I believe permanently fixed by a later release that by default not allowing more than 1/2 of total ram to be used as ARC cache.
2) For a long while, after every reboot I would need to execute a "zfs mount -a" to get the device mounted. Again, recently updated to new release and have no problem.
3) I use ZFS as a root filesystem at work, and sometimes after a ZFS/Kernel update, initramfs cannot mount the filesystem. To fix, I boot another zfs capable OS (another HD or livecd), do a force import, and then an export, then it works fine.
The snapshotting and the zfs send/receive by themselves are worth it. The ability to use linux apps/desktop/supported hardware makes ZFS on Linux just simply excellent.
As I've said before, the S11 11/11 release had a serious bug that crippled ARC for some folks (it wasn't clear why&when this happens). Basically, the ARC would get cleared down to almost zero and a kernel flag set that prevents it from ever growing. It was a silly bug (now fixed by Oracle), but unless you have a support contract (big bux) or they release a new ISO, you're stuck.
http://mail.opensolaris.org/pipermail/zfs-discuss/2012-January/050664.html
be warned.
http://mail.opensolaris.org/pipermail/zfs-discuss/2012-January/050664.html
be warned.
I do have empirical data (experiments/observations), you can follow zfsonlinux discussion,
ZFSonlinux still is going on project. or orhter discussions
GPL does not always "big pay", I know some spare times for fun/hobby and curiosity that would lead to job.
btrfs is a big improvement? this is depent on which side your are looking, on Oracle? linux kernel community? user community? enterprise/company objections?.
for me as a regular guy, btrfs promises a big improvement but in real life. Oracle has their own politics on that
the main issue ZFS is non compliance open source license, where ZFSonlinux born.. (besides userspace ZFS where is not robust since running on userspace)
It was nice when funding was available for ZFSonlinux.
ZFSonlinux really needs hardening than socializing.
do you remember SSH (openBSD originally)? SSH on linux did not need socializing, many use SSH due on demands.
the question is. btrfs development lacking would give a big boost for zfsonlinux as I see recently.
I try btrfs and do snapshot. it is ok, but.... many holes comparing wth zfsonlinux
as now.it means, Oracle asks much money... what a waste.:|
totally agree!!!, ZFSonlinux is not ready for enterprize use. we will see...
not speculations on zfsonlinux, this is on going work. you can see zfsonlinux milestone.
it works without significant problems, try it by yourself
I do not think Oracle would bring ZFS to linux, mostly on porting and change opensource license. Oracle can make money on that.
Oracle buys Redhat? I do not see in near future
could you reveal your speculations that can be shared to us? please..