Date: Mon, 5 Feb 2001 08:52:20 -0800 (PST) From: Luigi Rizzo <rizzo@aciri.org> To: ishizuka@ish.org (Masachika ISHIZUKA) Cc: freebsd-net@FreeBSD.ORG Subject: Re: packet loss when 'ipfw pipe list' with dummynet and bridge Message-ID: <200102051652.f15GqK123560@iguana.aciri.org> In-Reply-To: <20010205220806S.ishizuka@onion.ish.org> from Masachika ISHIZUKA at "Feb 5, 2001 10: 8: 6 pm"
next in thread | previous in thread | raw e-mail | index | archive | help
> >>> A better approach would probably be to set a semaphore before > >>> starting, and release it at the end, and keep interrupts enabled > >> > >> As I set "net.inet.ip.dummynet.expire=0", if it will affect > >> only to ip addresses founded newly when a semaphore is introduced, > >> I'll be happy. > > > > not clear what you mean, but in any case that variable will not help > > solving the problem that you are having. > > net.inet.ip.dummynet.expire only controls when flow descriptors > > are deleted, but they are always deleted when the system runs out > > of space, and new ones are always created when necessary. > > I am happy if no loss happened for bridged packets. > I keep the system not run out and no flow descriptors is deleted > because I collect traffic stats for my ip address range. I can accept > that the flow descriptors just now created are not reported. but it is not like this. At the moment, irrespective of the sysctl variables, reading the stats will put the kernel in a critical section where network interrupts are ignored until the copy is complete. cheers luigi > -- > ishizuka@ish.org > > > To Unsubscribe: send mail to majordomo@FreeBSD.org > with "unsubscribe freebsd-net" in the body of the message > To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-net" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200102051652.f15GqK123560>