I'm trying to get resierfs set up on a partition but get the following error:
Any ideas why?
squid optimisation guide: http://blog.last.fm/2007/08/30/squid-optimization-guide

mkfs.reiserfs /dev/sda2
mkfs.reiserfs 3.6.21 (2009 www.namesys.com)

A pair of credits:
Vitaly Fertman wrote fsck for V3 and maintains the reiserfsprogs package now.
He wrote librepair, userspace plugins repair code, fsck for V4, and worked on
developing libreiser4 and userspace plugins with Umka.

Alexander Zarochentcev (zam) wrote the high low priority locking code, online
resizer for V3 and V4, online repacker for V4, block allocation code, and major
parts of the flush code, and maintains the transaction manager code. We give
him the stuff that we know will be hard to debug, or needs to be very cleanly
structured.


Stat of the device '/dev/sda2' failed.[root@vps8259 myscripts]# df
Filesystem 1K-blocks Used Available Use% Mounted on
/dev/simfs 83886080 6384736 77501344 8% /
none 1572864 4 1572860 1% /dev
[root@vps8259 myscripts]#

/dev/sda2 doesn't exist or is inaccessible

Do I need to make a partition in the existing drive?
What I would like to do is have a folder for the squid cache. I heard it was better to have its own partition for speedy IO but not too bothered.
Do I then mount the /var/spool/squid to that drive? That's the squid cache folder.

Hi Qwerty,

I would say that having it on a separate partition will not improve performance, having it on a separate disk will. I suppose you could improve performance a little bit if you create the partition as the first partition on the disk so it is written to the outer portion of the platter, that is how they use to do it with Unix and the swap partition. Not sure if it really matters anymore.

If you really want performance and this is just cache(not perm data) I would do a strip across two or more drives and dedicate that as the cache drive, then have a separate disk for the rest of your system drive.

--
Brett Littrell
CISSP, CCVP, CCSP, MCNE, Project+, MCTS(AD,Server 2008, Network/Config, Win7)

I've seen that advice, too. In fact, my squid box uses reiserfs for all its partitions, but this is primarily because it is quite old*...

@blittrell
The argument is that reiser3, in default form (well, provided you've got noatime, of course) is particularly fast when dealing with lots of small files. Guess what the squid cache has to deal with?

I don't know if I'd bother with any of this messing around with reiser, these days; it would be certainly worth considering, for a box with a high throughput, but I'd probably just try ext4 and see how it went, and, if there were problems, do some experiments, and, if there weren't just leave it at 'the default'.

Technically, you mount the target partition as whatever name that squid uses (I think that you can change the name in squid.conf, if you really want, but I'm not sure that there is any particularly good reason for doing that....it is just a name, right?)

* by 'quite old', I mean that ext3 still wasn't fully released at the time.