Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 16 Nov 2012 17:50:14 -0500
From:      Zaphod Beeblebrox <zbeeble@gmail.com>
To:        Steven Schlansker <stevenschlansker@gmail.com>
Cc:        freebsd-fs@freebsd.org, John <jwd@freebsd.org>
Subject:   Re: RHEL to FreeBSD file server
Message-ID:  <CACpH0McAiaQjs%2BfrZxecu%2B4Uri%2BkPjfZ4aerNOOm7yr5yeiN3g@mail.gmail.com>
In-Reply-To: <16B803FB-0964-4237-8F25-291470E7EFB5@gmail.com>
References:  <50A130B7.4080604@cse.yorku.ca> <20121113043409.GA70601@neutralgood.org> <alpine.GSO.2.01.1211131132110.14586@freddy.simplesystems.org> <50A2B95D.4000400@cse.yorku.ca> <50A2F804.3010009@freebsd.org> <20121115001840.GA27399@FreeBSD.org> <20121115102704.6657ee52@suse3> <CACpH0Me_-MvqCbn5TmG892zUBMieOq8cDWPvxU5zLYjdPPKwXQ@mail.gmail.com> <20121116091747.2c1bfc55@suse3> <CACpH0Mc_OHhyReQmJVN2Sj3uy4swj8xQ=Y_0ntJhwFKeho-H_A@mail.gmail.com> <16B803FB-0964-4237-8F25-291470E7EFB5@gmail.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, Nov 16, 2012 at 2:38 PM, Steven Schlansker
<stevenschlansker@gmail.com> wrote:
>
> On Nov 16, 2012, at 11:24 AM, Zaphod Beeblebrox <zbeeble@gmail.com> wrote:
>
>> to connect 24 (or 30) drives.  If you're using "green" 2T's, the speed
>> of 4 disks on one channel is about half of the speed of
>> 1-disk-per-channel.
>> <snip>
>>
>> The reason I say all this... is that this config runs about $3500-ish
>> here in Canada (where green 2T's are ~$109).  The ZeusRAM drive up
>> there is 2/3 of that.
>
> Curious -- have you been running this setup for any length of time?  There's
> a fair number of horror stories about the "green" drives in particular.
>
> The power management is very aggressive about spin down, causing many unneeded
> power on/off cycles, dramatically reducing lifespan in a RAID configuration.
>
> Additionally, supposedly the error recovery is inappropriate leading to drive
> failure events.  (I believe the feature is known as TLER, time-limited error recovery)
>
> Have you run into this?

I do see the power down events.  If they array is quiet for some time,
it takes 10-ish seconds to respond... but then it's supposedly saving
power.  I've been running about a dozen of these setups for myself and
clients... the longest has been running since the early days of ZFS on
FreeBSD (my home array).  I find each array looses about 1 drive a
year.  If occasional errors pop up, they almost always indicate that a
drive _will_ fail.  Smart seems rather universally dumb on this
issue... at least for the green drives.

Due to the limitation of the port multipliers, increasing the speed of
this setup is often quite expensive (requiring RAID cards and 8x slots
and whatnot).  Without changing the drives, 24 ports of SATA cost
roughly $1200, last I looked at it.

While the majority of these are "backup" or "archive" file servers,
some do run production loads.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CACpH0McAiaQjs%2BfrZxecu%2B4Uri%2BkPjfZ4aerNOOm7yr5yeiN3g>