[OpenAFS] Server disk operations speed

Staffan Hämälä sh@ltu.se
Tue, 16 Apr 2013 07:46:49 +0200

On 2013-04-08 23:37, Markus Koeberl wrote:
>  > > Check the I/O scheduler settings on your virtual machine and on the
>  > > server.
>  > > The default cfq is not the fastest for this usecase. Especially if you
>  > > are using a battery backed hw raid you may test both set to noop.
> You can get it with:
> cat /sys/block/{DEVICE-NAME}/queue/scheduler

Ok, this is interesting. We've recently started using virtual 
fileservers, but have not done any changes to the IO scheduler. Is this 
the recommended setup for openafs fileservers on vmware? I just checked 
one of our virtual fileservers:
# cat /sys/block/sde/queue/scheduler
noop anticipatory deadline [cfq]

Btw, we mount the disks as "raw devices" in vmware, as this gives the 
best performance.

> Also check if you have disabled atime for the /vicepX filesytem

I know the noatime option can give better performance, but can it be 
used on /vicepX partitions? And will we get a performance benefit using it?