Subject: Re: possible kernel bug: stat mclpl after heavy net load
To: None <port-sparc64@NetBSD.org, current-users@NetBSD.org>
From: Pavel Cahyna <firstname.lastname@example.org>
Date: 12/16/2005 19:05:27
On Thu, Dec 15, 2005 at 02:24:36PM +0100, Stephan Pietzko wrote:
> Bruce O'Neel <email@example.com> wrote
> > I see a very similiar thing, but with nfsd rather than httpd, and on an
> > x86 system running current from this past weekend. And it's after
> > very heavy nfs load.
> Than it would be a general kernel-bug in the network memory pool?!
> How can i debug the problem or get more information about it?
> For now i wait for the next crash. The machine served 280G till now,
> serves 8-10MB/s and has 107 http-sessions oben.
> Every 10min i make a vmstat -m and grep mbpl and mclpl and a netstat
> into a file.
> What else can be helpfull for debugging?
compile a kernel with 'options SOSEND_NO_LOAN' ?