[quagga-dev 264] Re: [quagga-users 491] Re: Hardware Spec for a route server

Paul Jakma paul at clubi.ie
Mon Sep 29 13:14:28 BST 2003


On Mon, 29 Sep 2003, Gilad Arnold wrote:

> No it doesn't use it right now, for good reasons: work is generally
> synchronous in zebra, 

yes, but thats bad :)

> wrt to kernel management, namely zebra expects that a returned
> kernel call has accomplished its action 

yep.

> (that's the main reason for using blocking read calls to get
> netlink acks, an otherwise intolerable practice in zebra).

indeed - but what does zebra do currently if it gets an error back 
from the kernel, or if it gets nothing? (hmmm.. must get something).

> I assume that, in the presence of kernel work queues (somewhere
> around Q2-04... ;-> ),

2.6?

> this could be implemented, but then again -- only for netlink
> interface (is it worthwhile? not sure)

if nothing else it would save a wee bit on fewer context-switches for 
netlink comms. and if kernel is smart, it could save it work too. bit 
i dont know either.

> Gilad

regards,
-- 
Paul Jakma	paul at clubi.ie	paul at jakma.org	Key ID: 64A2FF6A
	warning: do not ever send email to spam at dishone.st
Fortune:
A statesman is a politician who's been dead 10 or 15 years.
		-- Harry S. Truman



More information about the Quagga-dev mailing list