tuning xsf

Issues Related to Hardware (Sound, Video, Printers, Scanners, Webcams, Touchpads, Keyboards, Sensors, Drives, GRUB, etc.)

Moderator: Moderators

Post Reply
Baby Hen
Posts: 6
Joined: Fri Oct 19, 2007 12:04

tuning xsf

Post by carpman » Tue Jan 15, 2008 11:08

Hello, been reading documents but still not sure what are best option to
use when creating xfs on my raid system. Raid card details below.

I have 2 raid 1 arrays which will have xfs partitions:

Code: Select all

Array 1 55gb with 1 40gb xfs partition for /home
Array set as stripe 64kb block 512 bytes

Code: Select all

Array 2 645gb 1 xfs partition using all space for file storage
Array set as stripe 128kb block 512 bytes

I have been looking at docs and come with following to create FS on 645gb
xfs partition.

Code: Select all

mkfs.xfs -f -d  agcount=42 -l size=64m /dev/sde5

The default agcount= was 32 but am not 100% sure how to work this out for
correct number?

I have been told that i should use agcount=8

I would mount with following options:

Code: Select all

mount -o logbufs=8,noatime,nodiratime

The other setting am not sure how to get right is this bit from docs: "XFS
allows you to select the stripe unit for a RAID device or stripe volume."

Now i know i need to set -d su= -d sw= but best to work this out for my

Also i have read that XFS's block size can not go above the page size of the running kernel. In linux this is 4k: is this correct?

The 645gb array/partition will be used for files 3mb to 250mb with most
being about 20/30mb

Raid is Areca 1210 raid controller:

Adapter Architecture

ARC-1210 uses IntelR IOP332 I/O processor. ARC-1220/1230/1260 uses
IOP333 I/O processor
PCI-Express X8 bus
256MB on-board DDR333 SDRAM with ECC protection(4/8-port)
One SODIMM socket with default 256MB of DDR333 SDRAM with ECC
protection, Upgrade to 1GB. An ECC or non-ECC SDRAM module using X8 or
X16 chip organization (12/16/24-port)
Write-through or write-back cache

Post Reply