[OpenAFS] improving cache partition performance

Edgecombe, Jason jwedgeco@uncc.edu
Mon, 29 Aug 2011 14:22:12 +0000

% fs getcacheparms
AFS using 145647975 of the cache's available 212484585 1K byte blocks.

I'm already using RAID 0, so I'm already striping. I'll have to do some ben=
chmarking with ext2.

Our web servers handle PHP web apps and large movie files.


Jason Edgecombe | Linux and Solaris Administrator
UNC Charlotte | The William States Lee College of Engineering
9201 University City Blvd. | Charlotte, NC 28223-0001
Phone: 704-687-3514
jwedgeco@uncc.edu | http://coe.uncc.edu | =A0Facebook
If you are not the intended recipient of this transmission or a person resp=
onsible for delivering it to the intended recipient, any disclosure, copyin=
g, distribution, or other use of any of the information in this transmissio=
n is strictly prohibited. If you have received this transmission in error, =
please notify me immediately by reply e-mail or by telephone at 704-687-351=
4.=A0 Thank you.

-----Original Message-----
From: openafs-info-admin@openafs.org [mailto:openafs-info-admin@openafs.org=
] On Behalf Of Harald Barth
Sent: Monday, August 29, 2011 8:13 AM
To: openafs-info@openafs.org
Subject: Re: [OpenAFS] improving cache partition performance

> For starters, I'd use ext2 instead of ext3, or perhaps use ext4 with
> journaling turned off.

I'd second the use of ext2 for AFS cache because you can allways wipe
and start over with an empty cache partition. During a crash, the
transactions that were not written to the server yet are lost anyway.

How big a cache do you need for your web server? Maybe stripe/mirror
would do much better than RAID5.

OpenAFS-info mailing list