Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 08 Jul 2008 21:50:18 +0200
From:      Kris Kennaway <kris@FreeBSD.org>
To:        mike <mike503@gmail.com>
Cc:        freebsd-fs@freebsd.org
Subject:   Re: Thinking of using ZFS/FBSD for a backup system
Message-ID:  <4873C4FA.2020004@FreeBSD.org>
In-Reply-To: <bd9320b30807080131j5e0e02a4y3231d7bfa1738517@mail.gmail.com>
References:  <bd9320b30807072315x105cf058tf9f952f0f5bb2a6a@mail.gmail.com>	<20080708100701.57031cda@twoflower.in.publishing.hu> <bd9320b30807080131j5e0e02a4y3231d7bfa1738517@mail.gmail.com>

next in thread | previous in thread | raw e-mail | index | archive | help
mike wrote:
> On 7/8/08, CZUCZY Gergely <phoemix@harmless.hu> wrote:
> 
>> Regardless of this, the system worked quite well. If ZFS were stable, this
>> easily could be our backup system. ZFS is great, awesome, but a bit unreliable
>> on FreeBSD, still needs some work.
> 
> Really? I thought ZFS for basic things was not too bad in FBSD now.
> 
> By basic I mean simple filesystem creation, snapshots and normal
> devices. Not some crazy SAN LUNs and weird volume management stuff.
> 
> I would really love to use FBSD as opposed to a Solaris derivative,
> since I know nothing about them and I'd have to dedicate a machine for
> it at home. Hrm. I wonder if I could just get by running a Solaris
> derivative inside of a VM in VMware or something.

ZFS needs careful memory tuning, but really, it's otherwise stable and 
it can be done.

(ports-i386:~>sysctl hw.ncpu
hw.ncpu: 4
(ports-i386:~)> sysctl hw.physmem
hw.physmem: 4275478528
(ports-i386:~)> uname -a
FreeBSD pointyhat.freebsd.org 8.0-CURRENT FreeBSD 8.0-CURRENT #31: Wed 
Jun 25 19:40:40 UTC 2008 
kris@pointyhat.freebsd.org:/usr/src/sys.cvs/amd64/compile/POINTYHAT  amd64
(ports-i386:~)> cat /boot/loader.conf
vfs.zfs.prefetch_disable=1
vm.kmem_size=1572864000

This machine is highly disk loaded, with 1.08TB of disk, a load average 
usually between 8-30, currently hosting 94 ZFS filesystems, 898 
snapshots, and making heavy use of ZFS features like cloning, 
incremental snapshot send/receive, etc.  The disk workload is highly 
vnode-intensive, involving concurrent rsyncs over trees containing 
hundreds of thousands of files, busy NFS exports to about 40 clients, 
cvs updates, etc, constantly cycling through millions of vnodes.

It works just fine.

Kris



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4873C4FA.2020004>