On Thu, Mar 09, 2000 at 06:23:03PM -0500, Tom Haapanen wrote:

 > Is there any likelyhood that this can be addressed in some future version of
 > NetBSD?  I actually checked, and another NetBSD/Alpha box here (which does
 > DNS and sendmail) is running with 32768 -- before that it wasn't stable.

Well, what do you mean "addressed"?

On the Alpha port, you can change this at run-time using the
kern.mbuf.nmbclusters sysctl variable.  This is specific to only
a few ports (the ones that use direct-mapped pool pages).

On other platforms, this isn't possible.

Luckily, you're using an architecture which doesn't fundamentally suck :-)

 > Is it not feasible for the kernel to recover from this, and get the network
 > running again?  Is this something inherent in the [Net]BSD kernel
 > architecture that cannot be cured?

Well, the kernel *DOES* recover... when it runs out, it attempts to free
any that are laying around in reassembly queues, etc. (for protocols
which will retransmit their data, etc.)

You don't, however, just want to grow the limit automatically... can you
say "denial of service"?

-- 
        -- Jason R. Thorpe <thorpej@shagadelic.org>