Date: Mon, 15 Feb 2016 18:28:08 -0800 From: Freddie Cash <fjwcash@gmail.com> To: Paul Kraus <paul@kraus-haus.org> Cc: Gary Palmer <gpalmer@freebsd.org>, FreeBSD Filesystems <freebsd-fs@freebsd.org> Subject: Re: Hours of tiny transfers at the end of a ZFS resilver? Message-ID: <CAOjFWZ5GaNcw1p92XPvsneZAW1NXhaN4qBuiQeLB5U2EcUyeLg@mail.gmail.com> In-Reply-To: <AF528DBA-CDBE-4ACB-A6FE-F5F87328EC08@kraus-haus.org> References: <8E04E52A-2635-4253-8140-F69495D7D0A6@panasas.com> <56C23E5B.7060207@multiplay.co.uk> <20160215213506.GB28757@in-addr.com> <AF528DBA-CDBE-4ACB-A6FE-F5F87328EC08@kraus-haus.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Feb 15, 2016 5:30 PM, "Paul Kraus" <paul@kraus-haus.org> wrote: > > On Feb 15, 2016, at 16:35, Gary Palmer <gpalmer@freebsd.org> wrote: > > > It should be noted that ZFS can do the right thing only at pool creation > > time. Once the pool has been created the sector size of the underlying > > disks is baked in and can only be changed by creating a new pool on > > the advanced format disks (or forcing the larger ashift value when > > you initially create the pool, even if the disks are really 512 byte > > sector drives) > > Is it baked in at the pool layer or the vdev layer ? I thought the ashift was set on a vdev by vdev basis. ashift property is set per vdev, and is set when the vdev is created. You cab have multiple different ashift values in a single pool, although it may be detrimental to performance. You can see the ashift value via "zdb poolname | grep ashift" Cheers, Freddie
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAOjFWZ5GaNcw1p92XPvsneZAW1NXhaN4qBuiQeLB5U2EcUyeLg>