[Bug 275308] EN tracking issue for potential ZFS data corruption

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=275308

--- Comment #20 from Rob Norris <robn@despairlabs.com> ---
(In reply to Troels Just from comment #17)

> would it be beneficial to do "zpool create ... -o feature@block_cloning=disabled" ? Does that provide any benefit over the sysctl you mentioned?

Disabling the feature makes the pool unable to clone blocks at all when asked
to do so. Disabling the sysctl makes it impossible to ask at all. On the same
system its much of a muchness, but if you move pools to other systems, that may
make a difference - the pool feature state will move with the pool, but the
sysctl is bound to the host.

> Is [#15333 about ZFS on LUKS] is in any way applicable on FreeBSD when running ZFS on top of GELI?

No, totally unrelated.

(In reply to Anderson Guzman from comment #19)

> [2.2] dmu_buf_will_clone: fix race in transition back to NOFILL

Its the fix for the cloning lock bug I mentioned right at the start, but we
think there are still cloning bugs around. Cloning will remain disabled in
2.2.x for now.

-- 
You are receiving this mail because:
You are the assignee for the bug.