Date: Mon, 22 Sep 2014 19:29:19 +0100 From: "Steven Hartland" <killing@multiplay.co.uk> To: "Tim Gustafson" <tjg@ucsc.edu>, <freebsd-fs@freebsd.org> Subject: Re: ZFS Warning Since Upgrade to 10.0 Message-ID: <6EF866714C19452C86C1BC0C99539D00@multiplay.co.uk> References: <CAPyBAS4OENwbDuqoRygQJF=RyrJJVXFgpz6qaT8vch=FO2j%2Bkw@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
----- Original Message -----
From: "Tim Gustafson" <tjg@ucsc.edu>
To: <freebsd-fs@freebsd.org>
Sent: Monday, September 22, 2014 5:14 PM
Subject: ZFS Warning Since Upgrade to 10.0
>I recently upgraded a ZFS file server from 9.2 to 10.0 and then
> started getting this warning when I run zpool status:
>
> status: One or more devices are configured to use a non-native block
> size. Expect reduced performance.
> action: Replace affected devices with devices that support the
> configured block size, or migrate data to a properly configured pool.
>
> I Googled around a bit, and understand the warning, but I have a
> problem: that zpool is 135TB and I don't have 135TB of disks laying
> around, nor the controllers necessary to support an additional 135TB
> of disks, to migrate this zpool to a properly configured one, nor
> could I easily have the server off-line for the requisite time that
> would be required to transfer 100+ TB of data from one set of hard
> drives to another.
>
> So my questions are:
>
> How much will this sub-optimal configuration affect performance?
That depends on your disks, as native 4k drivers when you send a 512
write it has to perform a COW operation.
The only real way to tell is to compare the two in a test with your
setup.
> Does the upgrade to 10.0 represent a reduction in performance, or was
> the reduction in performance always there and just not reported? This
> server is used to store genome data, so performance is pretty
> important, but the users were happy with the performance when it was a
> 9.2 server.
The issue was always there, its just ZFS now reports the issue.
> If I convince the users to go through an upgrade process to fix this
> issue, how much of a boost in performance can they expect to see? If
> it's a 2% boost, I don't think I can get them to invest in the
> upgrade, but it it's a 100% boost, I'm pretty sure I can.
Impossible to say, you could test on a smaller install if you want
to be sure.
Regards
Steve
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?6EF866714C19452C86C1BC0C99539D00>