FreeBSD ZFS File-System Code To Be Re-Based Over ZFS On Linux

[aht0]

According to what they say ZOL does have significant more features than their current Illumos upstream.
It's not about fixes, but about development speed.

What I understood (after reading multiple forums and bunch of mailing list emails), it's not so much about ZoL having more features but the fact that main contributors to the Illumos upstream are migrating to ZoL, Illumos itself is migrating to ZoL and OpenZFS would be pretty much dead in the water with next to no further development going on. Also ZoL's pace of development is faster, due to larger amount of Linux devs.

Main contributor to OpenZFS is migrating because Illumos upstream takes too fucking long pulling in contributed changes. Now Illumos woke I guess.

From what I understand ZoL still lacks support for NFSv4 acl's, TRIM etc. OpenZFS even has native dataset encryption, it's being security-tested by iXSystems and is just not yet mainlined.
It's supposedly about jumping ship early enough to have more word and weight in ZoL development.

[aht0]

Maybe with this rebase they'll rename it ZFS on Linux on BSD or ZoLoB for short

It sure is going to be funny. "ZFS on Linux" which shall be used on Illumos Solaris and FreeBSD by default but can't be used on Linux itself without jumping through legal hula hoops first..

[JAYL]

Maybe with this rebase they'll rename it ZFS on Linux on BSD or ZoLoB for short

[aht0]

lol, so much for "zfs is stable" fairytales

You ignore the fact that shit locking/racing on Linux wouldn't definitely do it on BSD. No racy systemd, different init, much fewer contesting processes. It remains theoretical possibility though, so it would be stupid to ignore such cases happening on Linux. ZFS has been integral part of FreeBSD since release 7, which came out back in 2008. 10 years has been enough to root out most issues. So keep your bigoted shit talk to yourself

[pal666]

lol, so much for "zfs is stable" fairytales

While working through the git history of ZoL I have also discovered that many races and locking bugs have been fixed in ZoL and never made it back to Illumos and thus FreeBSD

[starshipeleven]

The name is right there in the article. However, it seems a bit like finger pointing in the mailing list, IMO.

Delphix is still one of the "diamond sponsors" of OpenZFS that is moving to a different implementation (still within OpenZFS, ZOL is still part of OpenZFS) though. http://www.open-zfs.org/wiki/Main_Page

And if as they say illumos is lagging behind, this is not really going to improve the situation.

[some_canuck]

I'm just going to leave this here... https://illumos.topicbox.com/groups/...22eb01a/wakeup

[AndyChow]

"because the main sponsor of ZFS development on Illumos has decided they migrate to Linux/ZOL, so the Solaris/Illumos ZFS codebase is basically going in life support."

Names please!

Linux is in such a momentum, nothing really can come close. It does have its growing pains, but the pace of development is fantastic.

The name is right there in the article. However, it seems a bit like finger pointing in the mailing list, IMO.

[mbello]

"because the main sponsor of ZFS development on Illumos has decided they migrate to Linux/ZOL, so the Solaris/Illumos ZFS codebase is basically going in life support."

Names please!

Linux is in such a momentum, nothing really can come close. It does have its growing pains, but the pace of development is fantastic.

[aht0]

This don't make much sense:
1.) SPL still exists but it is integrated on ZFS module now instead of separated
2.) SPL as far as i know still exists on BSD as quoted on the mailing list
"The sources for FreeBSD's ZFS support are currently taken directly
from Illumos with local ifdefs to support the peculiarities of FreeBSD
where the Solaris Portability Layer (SPL) shims fall short"
Remember they are doing this simply because is easier to work with ZOL team than rely on Open/Solaris team that Oracle pretty make sure was non-existant at this point.

I am pessimist and assume that switching to ZoL would mean swapping Solaris ABI (what you described as "SPL") against Linux ABI. After all, ZoL has so far been developed/ported exclusively with Linux in mind.
There are bound to be "Linux'isms" in ZoL, intentional or not. I am pretty sure ZoL devs would not fancy ton of #ifdef's in their code. Unless of course, FreeBSD has enough weight with it's ZFS current "market share" to bully them into it.

If I was to continue with that paranoid and pessimistic mental train of thought: Let's assume FreeBSD starts requiring it's present Linux 64bit translation layer for running ZoL.
While Solaris shim was sufficient for ZFS and not for anything else, through Linux ABI one has ability to actually run Linux ELF-binaries on FreeBSD - as long as these binaries use Linux syscalls implemented in FreeBSD's Linux ABI.

hence it doesn't add any attack surface that weren't there before

Quite the opposite. If loading up Linux64.ko is mandatory for running/using ZFS, it does mean better than just theoretical capability of running also malware targeted at Linux. That was my point about increased attack surface. Not all malware is guaranteed to fail because of being dependent on systemd, bash shell (which, while not default, could be present in system because some installed port also required bash as a dependency) or unimplemented Linux syscalls in FreeBSD's Linux translation layer. Linux ABI in FreeBSD is concurrent with CentOS 6.x (not sure in exact version without proper checking). It can run plethora of Linux software still. Why not malware?

I would not discard that in the future open iteration of solaris will actually switch to work with the ZoL/FreeBSD team as well instead of spent their time begging Oracle for scraps

Solaris has 2 different versions of ZFS. One from OpenSolaris-derived Illumos, another is Oracle's closed-source "official" ZFS. Former is being used in FreeBSD and nobody has begged Oracle for scraps.