From mboxrd@z Thu Jan 1 00:00:00 1970 From: Erik Hugne Subject: Re: [PATCH net-next 1/2] tipc: fix excessive network event logging Date: Fri, 23 Jan 2015 07:44:30 +0100 Message-ID: <20150123064430.GA26059@haze> References: <1421943032-29924-1-git-send-email-erik.hugne@ericsson.com> <1421947487.2702.1.camel@perches.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Cc: , , , , To: Joe Perches Return-path: Received: from sesbmg23.ericsson.net ([193.180.251.37]:52611 "EHLO sesbmg23.ericsson.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750996AbbAWGqP (ORCPT ); Fri, 23 Jan 2015 01:46:15 -0500 Content-Disposition: inline In-Reply-To: <1421947487.2702.1.camel@perches.com> Sender: netdev-owner@vger.kernel.org List-ID: On Thu, Jan 22, 2015 at 09:24:47AM -0800, Joe Perches wrote: > On Thu, 2015-01-22 at 17:10 +0100, erik.hugne@ericsson.com wrote: > > From: Erik Hugne > > > > If a large number of namespaces is spawned on a node and TIPC is > > enabled in each of these, the excessive printk tracing of network > > events will cause the system to grind down to a near halt. > > The traces are still of debug value, so instead of removing them > > completely we fix it by changing the link state and node availability > > logging debug traces. > > Maybe some of these should be net__ratelimited(fmt, ...) We proposed that initially, but changed all to pr_debug after David's comment: http://www.spinics.net/lists/netdev/msg312902.html The topology information (links going up/down) can be accessed both via netlink ('tipc link list' command) and the topology server via a subscription API. //E