Is there any work going on to have ZFS as part of the default installation?

I guess one of the main reasons (but not the only one), why people chose FreeBSD is its great support for ZFS. It is a pity that ZFS on root is not part of the default installation. There are some great howtos there, but they look quite scary for us mortals. PC-BSD provides an easy installation of FreeBSD using ZFS, but am not sure how long that will last for. Is there any plan to make ZFS part of installation for the future releases? And for the moment is there a script that the faint hearted can use during installation?

Thanks
 
phoenix said:
ZFS installs have been possible since 9.0 via bsdinstall.

Sure but you'll have to use shell to set up ZFS. The guy is asking why doesn't bsdinstall have that feature built-in without having to do everything manually. I am wondering about this too and please don't mention PC-BSD because they're a different distro. I think FreeBSD can do better than this by adding an ZFS option to bsdinstall menu and that's not difficult to do. bsdinstall/sysinstall hasn't really changed that much since I used FreeBSD 4 and its still in text mode.
 
Would like to contribute and provide code to bsdinstall to have support for what you're asking?

And no I'm not being sarcastic or anything like that here. FreeBSD is a user contributed OS and everyone is welcome to contribute whatever it is that is desired.
 
blazingice said:
I guess one of the main reasons (but not the only one), why people chose FreeBSD is its great support for ZFS. It is a pity that ZFS on root is not part of the default installation. There are some great howtos there, but they look quite scary for us mortals. PC-BSD provides an easy installation of FreeBSD using ZFS, but am not sure how long that will last for. Is there any plan to make ZFS part of installation for the future releases? And for the moment is there a script that the faint hearted can use during installation?

Thanks

I use this ZFS script in shell during installation:
http://iceblood.googlecode.com/svn/FreeBSD_ZFS/freebsd_zfs_inst.sh
 
The PC-BSD installer can be used to install a completely standard FreeBSD install - It's one of the choices when selecting which type of environment to install. They've actually done a really good job of the installer although I can appreciate users wanting it in bsdinstall.

I've never used the PC-BSD installer to set up a basic FreeBSD machine myself and we really should have a ZFS option in bsdinstall. I would hope the devs are already looking at ways to built it in.

Personally, I've got used to doing the install by hand now, following the boot environments layout, which probably allows more freedom for customisation of the disk partitioning / pool layout than you'll get in an installer. (I believe PC-BSD are also using the boot environments layout in this or the next release and are building a GUI to manage them. They also provide an advanced disk configuration method which basically allows you to set the disks up by hand and return to the install, similar to what bsdinstall allows, but a bit more polished)
 
kpa said:
And no I'm not being sarcastic or anything like that here. FreeBSD is a user contributed OS and everyone is welcome to contribute whatever it is that is desired.

I would beg to differ. There are other distros such as GhostBSD, PC-BSD, etc. and why haven't they contributed to FreeBSD. Maybe they had and was rejected. I don't know but I get the feeling that FreeBSD Foundation somehow is limiting the FreeBSD potential to innovate and grow but I hope I'm wrong on this. Also FreeBSD Foundation have some programmers on staff as well.
 
As far as I know they have no programmers hired to do work on FreeBSD itself but only on the support infrastructure. All the work I've seen done while I've been following FreeBSD has been done by developers and users that have no direct ties to the FreeBSD foundation.
 
There are currently 2 members of staff hired by the FreeBSD Foundation to work on FreeBSD, although it's only very recent. One hired in March working on 'unmapped i/o' and another within the last month who's currently working on the kernel iSCSI target.

I appreciate the 'everyone can contribute' view but most users like myself (I have at least 30 FreeBSD servers), do not have the technical background to actually help write the OS. I think it benefits everyone if FreeBSD are aware of what users would like. Whether the project leads decide it's worth trying to get devs to work on it is another matter. I don't think ZFS in the installer is a high priority (especially as most users will want to customize the disks/pool manually anyway) but it should probably be on the roadmap somewhere.

Of course due to the way FreeBSD is developed and used, a lot of the development is corporate backed - The unmapped i/o mentioned earlier appears to be heavily involved with Netflix (probably because it'll make big improvements to their cache servers) and the bhyve project is backed by Netapp. I've no doubt someone has a vested interest in the iSCSI target as well (iSCSI target is a fairly key feature for a lot of server/virtualisation tech at the moment and FreeBSD is fairly useless at it) but even though I'm interested in that project, I can find little details about it or how it's progressing.
 
I understand why a default zfs option in bsdinstall doesn't have a high priority for companies who can employ skilled administrators. However, in my opinion, it would be a strategic investment to raise the popularity of FreeBSD among the unix/linux community. Bigger community can bring in more developers.

I remember when OpenSolaris released its time slider snapshot based on zfs. OpenSolaris joined the race quite late and not many people were paying attention to it. I don't think there was a great need for a time slider from a corporate perspective, but it was this killer feature that tempted many people to try OpenSolaris. Unfortunately, because of Oracle, it lost the momentum.

A great zfs experience out of the box could have the same effect. Corporations who provide funds for the development of FreeBSD should consider these kind of investments as advocacy for the future of its community.

I would love to help in this, but unfortunately I can hardly write a simple shell script. Will it not be easer to adopt what the team of PC-BSD has developed? I wonder how much interaction and code exchange there is between FreeBSD these projects?
 
Back
Top