[Samba] Samba on top of NFS question

[Fermin Molina]

On Mon, 2006-02-27 at 19:11 -0500, Ryan Taylor wrote:
> Thank you for the reply, its nice to know someone else is in the same boat.
> 
> Unfortunately our servers are also in production and therefore we can't try
> the patch.  We also have not found a good way to reproduce without throwing
> all the users on the system.
> 
> One interesting fact I have noticed since we have moved our data to the
> local samba machine: netstat -s -u returns
> # netstat -s -u
> Udp:
>     50898 packets received
>     16 packets to unknown port received.
>     0 packet receive errors
>     17163 packets sent
> #
> Where as before when on NFS we had 50%-75% of the number of "packets
> received" in "packet receive errors"


Mmm, it's weird.


> I am beginning to believe you are right, NFS cannot handle the locking
> required by Samba?


Please, read the thread "hanging smbd(s) revisited". I think is related
to our problems too.


> I am confused in two directions.  We use Whitebox4 and have noticed a
> newer version of nfs.utils.el4 rpm released but requires (because of two
> many dependencies) for use to upgrade to CentOS4... 2) We are using AMD X2's
> so are running the SMP kernel as well.  Could it be just a SMP issue? Which
> is why I am curious if anyone else had samba working in an environment
> similar to ours.


Maybe SMP could affect (I have 4 processors and linux SMP). But in the
thread I commented before, people have the same problems with different
platforms and configuration. I'm confused. I will wait for some answers
in the commented thread.

Cheers,

-- 
Fermin Molina Ibarz
Tècnic sistemes - ASIC
Universitat de Lleida
Tel: +34 973 702151
GPG: 0x060F857A