[Top] [All Lists]

Re: Pre-production questions

To: Joshua Baker-LePain <jlb17@xxxxxxxx>
Subject: Re: Pre-production questions
From: "Martin K. Petersen" <mkp@xxxxxxxxxxxxx>
Date: 20 Mar 2001 15:21:34 -0500
Cc: Linux xfs mailing list <linux-xfs@xxxxxxxxxxx>
In-reply-to: <Pine.LNX.4.30.0103201256210.2796-100000@xxxxxxxxxxxxxxxxxx>
Organization: Linuxcare, Inc.
References: <Pine.LNX.4.30.0103201256210.2796-100000@xxxxxxxxxxxxxxxxxx>
Sender: owner-linux-xfs@xxxxxxxxxxx
User-agent: Gnus/5.0808 (Gnus v5.8.8) XEmacs/21.1 (Canyonlands)
>>>>> "Joshua" == Joshua Baker-LePain <jlb17@xxxxxxxx> writes:

Joshua> The unit's default is a stripe size of 64 512 byte blocks.
Joshua> The stripe size can be set between 4 and 128 (incrementally,
Joshua> of course).  For the defaults, would this be correct:

Joshua> mkfs.xfs -d sunit=1,swidth=64

sunit = stripe width / number of disks

So mkfs.xfs -d sunit=8,swidth=64 for your 8 disks.

Joshua> Also, should I make the logfile any bigger than the default
Joshua> due to the size of the partition?

Depends how log-heavy the fs will be.  But in general we tend to use a
bigger log than the 1000b default.

We've talked about adding heuristics to mkfs to make a qualified
guess.  That's on my todo list.

Joshua> The system is maxed out at 128MB of cache RAM.  The primary
Joshua> use will be reading and writing (over NFS) lots of largish
Joshua> (250+MB) data files.

In that case I'd probably up the stripe size a bit.  But do some
benchmarks for yourself before you decide.  RAID controllers vary a

Martin K. Petersen, Principal Linux Consultant, Linuxcare, Inc.
mkp@xxxxxxxxxxxxx, http://www.linuxcare.com/
SGI XFS for Linux Developer, http://oss.sgi.com/projects/xfs/

<Prev in Thread] Current Thread [Next in Thread>