ZFS performance help sought

[Mason Loring Bliss]

On Thu, Jan 21, 2016 at 09:08:57PM -0800, David Christensen wrote:

> 1.  Have you tested the disks, cables, and controllers individually using
> disk manufacturer diagnostics?  E.g. for Seagate, SeaTools for DOS (bootable
> CD based on FreeDOS; MS-DOS not required):

Yeah. Plus, this only happens with FreeBSD, and I don't believe that Linux is
quietly working around some really subtle weirdnesses in my drives.

Honestly, the verification you're suggesting would be an absolutely
tremendous amount of work and pretty disruptive only to highlight that the
hardware is okay and FreeBSD or the way it uses ZFS is doing something odd. I
did the "here, let me test the devices" dance when I discovered an OS-
agnostic flaw in OpenZFS, and I'm just not going to do it again. I don't want
to seem ungrateful for your time in writing up the questions, but I don't see
how I can reasonably obtain the data you want. It'd be much less work to just
go to a known-good Linux configuration, but that doesn't make FreeBSD better,
which is a giant part of my goal in having switched back.

The iostat output earlier in the thread showed a snapshot of relative
performance, and I'm happy to do more diagnostics that don't involve
disassembling this particular box.

-- 
Mason Loring Bliss             mason at blisses.org            Ewige Blumenkraft!
(if awake 'sleep (aref #(sleep dream) (random 2))) -- Hamlet, Act III, Scene I