Announcement

Collapse
No announcement yet.

Linus Torvalds Doesn't Recommend Using ZFS On Linux

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Dedobot
    replied
    Booting the OS from desired snapshots is neat.
    Definitely very useful for desktops and mobile devices. I have no idea if this can be implemented w/o zfs code in the kernel ?
    Zfs boot environments works wonderful in Solaris, i know FreeBSD have that option too ,

    Leave a comment:


  • k1e0x
    replied
    Originally posted by gilboa View Post
    Ignoring for a second the obvious issues with CDDL vs. GPL and Oracle's horrible litigious nature (both of which makes ZOL {ZFS on Linux} legally banned in any real enterprise environment, unless that is, you are using an Oracle licensed OS), one must wonder if any of the many pro-ZOL users, really attempted to deploy it in a real, honest to God, large scale production environment (E.g. GlusterFS cluster over ZFS).

    Anyone? Cause at least in my experience, unlike ext4/EXFS, when ZOL breaks (and I've seen it break badly in production environment, including very recently) you get to keep all the pieces.

    - Gilboa
    I can assure you ZoL *IS* used in enterprise. I've used ZFS in enterprise since ~2006 on Solaris, FreeBSD and Linux. It is pretty much the fix for ransomware.

    The CDDL nor the GPL have restrictions on usage, only on distribution. You can and always have been able to do whatever you want on your own systems.

    I posted this before. 100 PB - ZoL Implementation. Storage of that size actually isn't that unique for ZFS (typically more Solaris and FreeBSD but ZoL is getting more popular.) Most enterprises don't share information on their storage platforms. Since that one is a quite impressive research system they do have details posted about it.
    https://medium.com/codedotgov/oss-sp...x-6596fca6e5f6

    "unlike ext4/EXFS, when ZOL breaks (and I've seen it break badly in production environment, including very recently) you get to keep all the pieces." -- seriously? You're going to make the claim ext4 is *better* at data integrity than ZFS? No.. It's not. Not in any universe. Like Linus you don't seem familiar with ZFS.. that is ok.. but unlike Linus you shouldn't go making claims when you don't have the basic facts.

    FYI, ZFS is a copy on write, always consistent file system with a checksum on every data block.
    EXT4 is... not. It is a journaled, extent based file system that updates blocks in place making it not always consistent. Do you like fsck? Or data blocks with silent corrupt files? EXT4 is for when you want a small but present possibility of your pictures looking like this and never knowing till you open it.
    Last edited by k1e0x; 01-20-2020, 04:33 PM.

    Leave a comment:


  • gilboa
    replied
    Ignoring for a second the obvious issues with CDDL vs. GPL and Oracle's horrible litigious nature (both of which makes ZOL {ZFS on Linux} legally banned in any real enterprise environment, unless that is, you are using an Oracle licensed OS), one must wonder if any of the many pro-ZOL users, really attempted to deploy it in a real, honest to God, large scale production environment (E.g. GlusterFS cluster over ZFS).

    Anyone? Cause at least in my experience, unlike ext4/EXFS, when ZOL breaks (and I've seen it break badly in production environment, including very recently) you get to keep all the pieces.

    - Gilboa

    Leave a comment:


  • ryao
    replied
    Originally posted by smitty3268 View Post

    Disagree.

    Obviously patents are how they'd attack the GPL code, but I think it's very clear that would be a tougher argument to make than the straightforward copyright claim they'd have with the ZFS code.

    In the end, it's just a matter of degree though. I agree they absolutely could drag the gpl code into court if they wanted to.
    They don’t have a copyright infringement claim to make according to attorneys.

    https://www.softwarefreedom.org/reso...rnel-cddl.html

    The only potential issue according to that would be all of the Linux copyright holders forming a consensus “to prefer the literal meaning to the equity of the license”.
    Last edited by ryao; 01-18-2020, 12:26 PM.

    Leave a comment:


  • smitty3268
    replied
    Originally posted by ryao View Post

    The GPLv2 has no patent grant, so code under it is far more vulnerable than code under the CDDL. Being under the GPLv2 does not make it more safe. If anything, it is less safe. That is why the GPLv3 was written.
    Disagree.

    Obviously patents are how they'd attack the GPL code, but I think it's very clear that would be a tougher argument to make than the straightforward copyright claim they'd have with the ZFS code.

    In the end, it's just a matter of degree though. I agree they absolutely could drag the gpl code into court if they wanted to.

    Leave a comment:


  • ryao
    replied
    Originally posted by nivedita View Post
    The CDDL still has the terms allowing Sun (now Oracle) to publish a new version, and for anyone to then choose to use that version. See section 4. There is no notice in the ZFS source specifically prohibiting use of newer versions. If Oracle says CDDL 2.0 is a copy of GPLv2, it's done.
    All new OpenZFS code has a CDDL version pin to prevent Oracle from being able to change the license terms.

    Leave a comment:


  • k1e0x
    replied
    Originally posted by nivedita View Post
    No-one denies that the two licenses are currently incompatible.
    I do. Other people do too. It's complicated. But you have to take into account GPLv2 clause 0 being void in the US and EU and the ability to comply with both licenses at the same time meaning one can follow the terms of both licenses and in doing so does not cause harm to either license. That is Canonical's opinion. (and they aren't the only ones, several university legal professors have agreed with them)

    The FSF and RedHat take a different opinion (however it is rumor privately there is some dissent in that camp and Stallman and his underlings have been in dispute about this)

    As read the licenses are incompatible. In actual effective practice.. maybe not.

    Regardless that is moot because it's not in tree, and OpenZFS doesn't really want to put it in tree even if Linus gave the ok. ZFS doesn't cater to only Linux, they support 5 OS's but they do want it to work well on Linux.

    Ars Technica picked this up btw.
    https://arstechnica.com/gadgets/2020...straight-dope/
    "Linus should avoid authoritative statements about projects he's unfamiliar with."

    I think they are being generous to Linus. "Don't worry people data integrity is a buzz word and it's not even as fast! You don't need to know your data is *right*, you'd loose 5-10% of your performance that way!"
    Last edited by k1e0x; 01-16-2020, 07:46 PM.

    Leave a comment:


  • nivedita
    replied
    Originally posted by k1e0x View Post
    And again sorry oiaohm.. Linux dm can't tell if data on one drive is correct from data on another as it has no checksum on the data. Corruption on one side of a mirror can't be compared to the other. All it can do is read. In ZFS's case it reads, compares the checksum and then can take action if it's wrong such as looking in the array for another copy of the data and even going back and fixing the bad block. That is what it means by self healing.
    This aspect should actually be emphasized a lot more as it is a crucial advantage of checksumming in the filesystem. I'd say more so than bitrot even. Having a mirror doesn't help if you don't know which side has the correct data any more. On regular mirrors or RAID if you ever get out of sync on which drive had the failure your entire array is basically gone.

    Eg, I've had a case where a botched Windows install overwrote part of one of the drives in my storage zpool. If that had been a regular RAID, I'd have had to sweat bullets to make sure I kept track of exactly which drive needed to be rebuilt. With raid-z I just booted back into linux and scrubbed the pool, and the filesystem handles everything.
    Last edited by nivedita; 01-16-2020, 06:45 PM. Reason: Added example

    Leave a comment:


  • nivedita
    replied
    Originally posted by oiaohm View Post

    https://web.archive.org/web/20060816...PL_redline.pdf
    the problem in 3.1 of CDDL is.
    be distributed only under the terms of this License.

    This is straight from MPL 1.1 and is in CDDL 1.0 as is. This means CDDL cannot be over-licensed with any license just like MPL 1.1 does not have to be GPL. Lets say at some point Oracle released a CDDL 1.1 you could not over-license CDDL 1.0 with CDDL 1.1 no matter how you worded CDDL 1.1. Yet MPL 1.1 could be over-license by newer version of license because it contained
    or a future version of this License released under Section 6.1

    This was removed from CDDL. So the issue CDDL is not simple to fix. The fix to CDDL is throw the complete code base away and rewrite from scratch. Everything tainted by CDDL is not fixable legally.

    That distributed only under terms of this License blocks the normal thing under copyright where you can licence your own work under as many licenses as you like. Yes even if you ask every developer in a CDDL work for permission to re-license and they said yes the CDDL license still directly forbids this. CDDL is many levels worse than GPL in this regard you can re-license a GPL work if you can get all the developers to agree. This is most likely why Oracle legal department does not answer on re-license or allow into Linux kernel because you most likely cannot ever legally. Yes CDDL is a viral license on a complete different level once you put something under CDDL its under CDDL forever and nothing else.
    The CDDL still has the terms allowing Sun (now Oracle) to publish a new version, and for anyone to then choose to use that version. See section 4. There is no notice in the ZFS source specifically prohibiting use of newer versions. If Oracle says CDDL 2.0 is a copy of GPLv2, it's done.

    You're also missing my point. No-one denies that the two licenses are currently incompatible. What is controversial is whether Sun specifically intended to make it incompatible with Linux, or it just turned out that way because they weren't really considering that (remember at the time they were just open-sourcing the Solaris operating system).

    Leave a comment:


  • k1e0x
    replied
    Yes, using a larger sector size and placing the checksum next to the block is bad because it can't protect from phantom read/write or write/read from wrong sector. ZFS solves this problem and can chain verify itself down entire tree of blocks.

    And again sorry oiaohm.. Linux dm can't tell if data on one drive is correct from data on another as it has no checksum on the data. Corruption on one side of a mirror can't be compared to the other. All it can do is read. In ZFS's case it reads, compares the checksum and then can take action if it's wrong such as looking in the array for another copy of the data and even going back and fixing the bad block. That is what it means by self healing.

    Linux DM has proposed adding this but it's logic for the proposal is pretty bonkers. They write the block out then wait 5 seconds for the compression to finish then calculate the checksum.. whoa.. no thanks. ZFS does it in ram before the write. (this is what I mean by shoehorn and duct tape on features.) In ZFS writing compressed data takes less time than uncompressed.. because well... you're writing less data! Simple. In Linux DM writing less data takes 5 seconds more time?? How does that logic work? Did it double write the data? How did they determine 5 seconds was right? it sounds like an ass pull number. Ick ick.. just no.. no.. thanks for trying.

    XFS has checksums also, but only on metadata. They cite it's too slow to do it on data blocks. (A problem ZFS's "terrible" design allowed them to solve.. if only XFS's design was as bad as ZFS's they could have data block checksums. )
    Last edited by k1e0x; 01-15-2020, 09:19 PM.

    Leave a comment:

Working...
X