> -----Original Message-----
> From: Sean Boudreau [mailto:community-noreply@qnx.com]
> Sent: Wednesday, December 16, 2009 1:16 PM
> To: general-networking
> Subject: Re: NMBCLUSER
> 
> On Wed, Dec 16, 2009 at 01:12:01PM -0500, Mario Charest wrote:
> >
> >
> > > -----Original Message-----
> > > From: Sean Boudreau [mailto:community-noreply@qnx.com]
> > > Sent: Wednesday, December 16, 2009 1:01 PM
> > > To: general-networking
> > > Subject: Re: NMBCLUSER
> > >
> > > On Wed, Dec 16, 2009 at 12:04:26PM -0500, Mario Charest wrote:
> > > > On occasion we have io-pkt-v4 that self terminate ( no dump file
> ).
> > > All I could see, is a message about increasing NMBCLUSTERS in
> sloginfo.
> > > I guess I could do that with sysctl kern.mbuf.nmbclusters but i'd
> like
> > > to understand what is this about?  Is increasing the value the real
> > > solution, what is the cause.
> > >
> > > io-pkt reached the kern.nbuf.nmbcluster limit.  It doesn't
> > > exit on that condition though; however things may limp
> > > along if packet buffers can't be allocated.
> >
> > Thanks Sean,
> >
> > Is this related to rx and tx descriptor?  I have 2048 rx and 2048 tx
> and there are two interfaces for a total of 8192 which is the number of
> nmbclusters available.
> 
> Yes, check out 'netstat -m'.

Thanks, maybe at startup or when interfaces are mounted, io-pkt-v4 should do something about having more descriptors 
then mbuf? 

> 
> Regards,
> 
> -seanb
> 
> 
> 
> _______________________________________________
> 
> General
> http://community.qnx.com/sf/go/post43953
>