On Sat, Sep 21, 2002 at 09:28:48PM +0200, Markus W Kilbinger wrote:
> >>>>> "Manuel" == Manuel Bouyer <bouyer@antioche.eu.org> writes:
> 
>     >> - How to avoid these hangs? Further NMBCLUSTERS increase?
> 
>     Manuel> No, it's not an NMBCLUSTERS issue. It's running out of
>     Manuel> mbuf headers. Does unplugging the network stop the message
>     Manuel> stream ?
> 
> > I didn't try this... maybe next time :-/
> 
> Now it's the fifth time such a hang up happens! Unplugging the network
> cable interrupts the message stream (ti: link down), which will be
> continued after replugging in the network cable (ti: link up). The
> hang up remains complete otherwise (-> reboot).
> 
> Is this a 'ti' specific problem, then?

Not sure. It can't allocate a mbuf, I suspect because there are already too
much allocated. But it may be something else eating the mbufs. Can you log in
and run netstat and vmstat to see the states of some data structures ?
Usefull things could be vmstat -m (make sure your kernel is compiled with
options KMEMSTAT), and netstat -m.

> 
> > Does anybody else see these messages?
> 
> Really nobody?

Well, my machine with a tigon is still running 1.5.2+, because of NFS server
bugs in 1.6 ...

-- 
Manuel Bouyer <bouyer@antioche.eu.org>
--