Accelerating 'mountd'

Accelerating 'mountd'

[Martin Spott]

Hello,
I'm having a setup with NFSv3 service over UDP from a 2.4.31 kernel.
Multiple diskless clients mount all their filesystems during boot and
the the crazy guy who contructed this diskless boot mimic lets each
client mount _ten_ NFS shares from the server (for some complicated
automatic configuration purposes).

In principle this is a stable setup but it comes to a bottleneck, when
twenty diskless client intend to mount their shares almost at the same
time. Therefore the server is supposed to handle about 200 mount
requests within ten seconds ....  but when I look at my server I
realize that at best only two requests are being handled per second.

In this context I'm seeking for answers to the following questions:
1.) Should a 'mountd' from mostly current nfs-utils be able to handle
    such a load on performant hardware with a sane setup ?
2.) Are there typical environments that degrade the performance of the
    'mountd' where I should have a closer look at ?
3.) Do I probably focus on the wrong culprit ....  ?

This sort of difficulties is a bit new to me and I know that such a
setup sounds a bit crazy, but I simply have to face the situation  :-)

I'd appreciate if you'd share some of your experience on this topic.

Thanks,
	Martin.
-- 
 Unix _IS_ user friendly - it's just selective about who its friends are !
--------------------------------------------------------------------------


-------------------------------------------------------
This SF.Net email is sponsored by:
Power Architecture Resource Center: Free content, downloads, discussions,
and more. http://solutions.newsforge.com/ibmarch.tmpl
_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs

Re: Accelerating 'mountd'

[Garrick Staples]

[-- Attachment #1: Type: text/plain, Size: 492 bytes --]

On Wed, Oct 19, 2005 at 04:09:21PM +0000, Martin Spott alleged:
> 2.) Are there typical environments that degrade the performance of the
>     'mountd' where I should have a closer look at ?

I'm far from an NFS expert, but I've always seen slow mounts come from
slow or missing DNS resolution.  Simply adding the clients to the
server's /etc/hosts and running nscd can often make a huge difference.

-- 
Garrick Staples, Linux/HPCC Administrator
University of Southern California

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: Accelerating 'mountd'

[Martin Spott]

Martin Spott wrote:

> In principle this is a stable setup but it comes to a bottleneck, when
> twenty diskless client intend to mount their shares almost at the same
> time. Therefore the server is supposed to handle about 200 mount
> requests within ten seconds ....  but when I look at my server I
> realize that at best only two requests are being handled per second.

Actually I had to realize that I've been fooled. A co-worker, although
having all NFS clients entered in /etc/hosts was claiming to face the
same problem - but this actually was not the case.
Now after entering all clients in /etc/hosts my server happily serves
20 mount requests per second - or even more.

Sorry for the noise,
	Martin.
-- 
 Unix _IS_ user friendly - it's just selective about who its friends are !
--------------------------------------------------------------------------


-------------------------------------------------------
This SF.Net email is sponsored by the JBoss Inc.
Get Certified Today * Register for a JBoss Training Course
Free Certification Exam for All Training Attendees Through End of 2005
Visit http://www.jboss.com/services/certification for more information
_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs