From mboxrd@z Thu Jan 1 00:00:00 1970 From: Patrick McHardy Subject: Re: [PATCH] netfilter: scheduled removal of CONFIG_NF_CT_ACCT Date: Mon, 31 May 2010 17:56:03 +0200 Message-ID: <4C03DC13.9080307@trash.net> References: <1275220311-12578-1-git-send-email-tobias.doerffel@gmail.com> <4C03D4BA.9070809@trash.net> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-15 Content-Transfer-Encoding: 7bit Cc: Tobias Doerffel , netfilter-devel@vger.kernel.org, Krzysztof Piotr Oledzki To: Jan Engelhardt Return-path: Received: from stinky.trash.net ([213.144.137.162]:60525 "EHLO stinky.trash.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751610Ab0EaP4H (ORCPT ); Mon, 31 May 2010 11:56:07 -0400 In-Reply-To: Sender: netfilter-devel-owner@vger.kernel.org List-ID: Jan Engelhardt wrote: > On Monday 2010-05-31 17:24, Patrick McHardy wrote: >> Tobias Doerffel wrote: >>> Removal of the CONFIG_NF_CT_ACCT was already scheduled for 2.6.29, >>> therefore remove it now and set nf_conntrack.acct depending on >>> CONFIG_NETFILTER_XT_MATCH_CONNBYTES. >> This has already been discussed multiple times. The connbytes >> match needs accounting, so it needs to enable it when the first >> rule containing a connbytes match is added. This needs to be >> done in the namespace where the rule is added. > > If it is done as late as rule insertion, connbytes might match (or > not - and thus be a problem) when there are still CTs around without > a counter tuple. Hm. Yeah, we also discussed that part, but we can't do any better. The alternative for people is to manually enable it when loading conntrack.