SlackwareThis Forum is for the discussion of Slackware Linux.
Notices
Welcome to LinuxQuestions.org, a friendly and active Linux Community.
You are currently viewing LQ as a guest. By joining our community you will have the ability to post topics, receive our newsletter, use the advanced search, subscribe to threads and access many other special features. Registration is quick, simple and absolutely free. Join our community today!
Note that registered members see fewer ads, and ContentLink is completely disabled once you log in.
If you have any problems with the registration process or your account login, please contact us. If you need to reset your password, click here.
Having a problem logging in? Please visit this page to clear all LQ-related cookies.
Get a virtual cloud desktop with the Linux distro that you want in less than five minutes with Shells! With over 10 pre-installed distros to choose from, the worry-free installation life is here! Whether you are a digital nomad or just looking for flexibility, Shells can put your Linux machine on the device that you want to use.
Exclusive for LQ members, get up to 45% off per month. Click here for more info.
I saw a SlackBuild for zfs-fuse, limited to 128MBytes, except in some circumstances, and when perusing zfsonlinux.org FAQs it had mentioned some technical drawbacks of implementing ZFS-FUSE instead of just ZFS.
I was wondering what pearls of wisdom I might derive here in the *home* forum, as I'm considering a sort of backblaze solution for a couple of clients.
I'm usually an XFS kinda guy, and like ZFS on Solaris but I'm considering bringing ZFS to Linux and not certain as to the drawbacks I might encounter w/a 135TB box if I use FUSE.
I was under the impression that ZFS liked to work with the raw disks. Now 135 TB, using say 3 TB drives, will work out to 45 drives. That's way past the normal capacity of the average server, without using special RAID cards. Often those RAID cards don't like to pass back the raw disk, but rather present as pseudo devices. Usually you're getting fancy hardware RAID services (raid10, raid5, raid6 etc), which you pay for. And the cards are normally optimised for that type of access. I'd be interested to see exactly how you'll be connecting this all up.
ZFS is a very nice file system to have if you can use it. It's far more featured and advanced that about any file system Linux has the current ability to offer. However, due to the finickiness of the GPL, the CDDL license and software created under it just seem to be incompatible, though most blame for that rests within both licenses bylaws and conscripts. These could easily be rectified with amendments to the GPL.
By all regards you COULD create a personal non-redistributable Linux based system that offers ZFS, but the main issue is, it could be considered an illegal Linux distribution and could not be redistributed.
Now 135 TB, using say 3 TB drives, will work out to 45 drives.
Yes that's precisely what it is. And although I was rather skeptical piling all of that into a single 4U case (I think I told my partner to bring the eggs and specify sunny side up or scrambled), the specs appear impressive for a low end cheap storage solution.
Quote:
Originally Posted by Mark Pettit
I'd be interested to see exactly how you'll be connecting this all up.
This is an idea for selling storage solutions that one of my partners wants me to investigate. Consequently, the funding won't be coming out of my pocket with the exception of rackspace in my private suite, so when he suggested I look at this 4U solution my interest was piqued.
Here's the pieces lined up all nice and neat, much of which I can get wholesale:
ZFS ...due to the finickiness of the GPL, the CDDL license and software created under it just seem to be incompatible, though most blame for that rests within both licenses bylaws and conscripts. These could easily be rectified with amendments to the GPL.
By all regards you COULD create a personal non-redistributable Linux based system that offers ZFS, but the main issue is, it could be considered an illegal Linux distribution and could not be redistributed.
And therein lies my dilemma, although I'll be racking this and we'll be selling space in some manner or another, so perhaps the distribution part isn't going to be a legal issue for me in the way it would restrict someone who intended to sell the hardware/software combination as a complete package.
That's why zfsonlinux.org suggested that one can get around this issue by using zfs-fuse, and even though the constraints you've pointed out may not be a concern given my particular deployment objective, I'm still wondering what the caveats concerning *technical issues* incorporating zfs-fuse in userspace are in comparison to just zfs.
Personally, I think packing that many SATA drives into a box and expecting performance is kind of nuts, but it's a chance to play with tinker toys and it's not going to cost me anything out of my pocket.
I am perfectly fine sticking w/my g5 Proliants w/SAS drives under OpenFiler and iSCSI, (I can get them cheap on eBay and the boiz at the NOC can swap out bad drives in two seconds when the lights change color) but if my buddy's going to fund scary monsters and let me hold the screwdriver I'm game to play
Y'know @jefro, That's a great idea for me to look into. Deb's actually my second fav Linux distro and I've been meaning to check out some of their other arch's for some time.
Other than that I was also considering FreeBSD w/ZFS or Dragonfly w/HAMMER. Btrfs is nice but might not yet be safe enough to entrust customer data to.
I've used btrfs for years. All my systems run (at least) the root and home partitions. But I don't deal with production/(real/valuable) user data.
And I am (and always have been) anal about backups.
I thought I saw something about Oracle (unbreakable maybe) looking at btrfs as the default. So it must be stabilizing somewhat.
I was able to consistently break the entire filesystem by removing a drive from a RAID10 test setup and trying to initiate a rebuild after replacement. That was a bit flaky at the time, but was at least two years ago.
Fedora have been making noises about making it the default for F16 and then F17, but never did I think (haven't checked F17).
I've used btrfs for years. All my systems run (at least) the root and home partitions. But I don't deal with production/(real/valuable) user data.
And I am (and always have been) anal about backups.
I thought I saw something about Oracle (unbreakable maybe) looking at btrfs as the default. So it must be stabilizing somewhat.
I was able to consistently break the entire filesystem by removing a drive from a RAID10 test setup and trying to initiate a rebuild after replacement. That was a bit flaky at the time, but was at least two years ago.
Fedora have been making noises about making it the default for F16 and then F17, but never did I think (haven't checked F17).
f17 is ext4 as default. if sun is planning to adopt btrfs as you say then it must be good considering they use zfs as default which is already more then good enough.
Oracle uses Linux for some of their servers so it's probably why they are looking into BtrFS.
BtrFS is a nice robust filesystem but it does have it's flaws such as you can not boot a BtrFS partitioned drive using conventional bootloaders like LILO. Because many people partition drives sometimes with only a single root partition / and a swap partition /swap it's best to avoid BtrFS even if it's a more robust system which is problematic.
As far as ZFS integrating with Slackware... honestly I couldn't tell you because no efforts have ever been attempted to write a Slackware specific native ZFS supporting kernel and install-time or post-install partitioning tools.
Plus I have no idea if ZFS can be booted with LILO also, as it and file systems like XFS, BtrFS, and such may require using GRUB which Slackware does not include.
LinuxQuestions.org is looking for people interested in writing
Editorials, Articles, Reviews, and more. If you'd like to contribute
content, let us know.