[Beowulf] diskless cluster nfs

[Craig Tierney]

On Tue, 2004-12-07 at 15:23, Josh Kayse wrote:
> Ok, my first post, so please be gentle.
> 
> I've recently been tasked to build a diskless cluster for one of our
> engineers.  This was easy because we already had an image for the set
> of machines.  Once we started testing, the performance was very poor. 

What performance is poor?  Is it the whole code that is
slowing down or is it just the disk IO?


> Basic setup follows:
> 
> Master node: system drive is 1 36GB SCSI drive
>                      /home raid5 5x 36GB SCSI drives

Have you tuned the performance of your raid5 device?
Depending on your controller, should probably be seeing
100 MB/s for both read and write.  Or is this software raid?




> Master node exports /tftpboot/192.168.1.x for the nodes.
> 
> all of the nodes are diskless and get their system from the master
> node over gigabit ethernet.
>  All that worsk fine.
> 
> The engineers use files over nfs for message passing, and no, they

So this used to work ok?  Did it work with the same NFS server
before you had it export the diskless filesystem?


> will not change their code to mpi even though it would be an
> improvement in terms of manageability and probably performance.
> 
> Basically, my question is:  what are some ways of testing the
> performance of nfs ande then, how can I improve the performance?
> 
> Thanks for any help in advance.
> 
> PS: nfs mount options: async,rsize=8192,wsize=8192,hard
>        file sizes: approx 2MB

If possible, increase the MTU on your interfaces.

Also test your NFS performance and compare it to the
raw disk performance.

Craig