From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from sog-mx-4.v43.ch3.sourceforge.com ([172.29.43.194] helo=mx.sourceforge.net) by sfs-ml-2.v29.ch3.sourceforge.com with esmtp (Exim 4.76) (envelope-from ) id 1WIIYi-0007Pk-Ek for ltp-list@lists.sourceforge.net; Tue, 25 Feb 2014 14:00:32 +0000 Date: Tue, 25 Feb 2014 15:00:17 +0100 From: chrubis@suse.cz Message-ID: <20140225140017.GC14376@rei> References: <20140211142336.GA14876@rei.Home> <1392131944-27263-1-git-send-email-stanislav.kholmanskikh@oracle.com> <1392131944-27263-2-git-send-email-stanislav.kholmanskikh@oracle.com> <20140221181117.GA15723@rei> <530C5E99.6050809@oracle.com> <20140225122538.GA12159@rei.Home> <530CA0A3.403@oracle.com> MIME-Version: 1.0 Content-Disposition: inline In-Reply-To: <530CA0A3.403@oracle.com> Subject: Re: [LTP] [PATCH V2 2/2] msgctl11: process message queues by portions List-Id: Linux Test Project General Discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Errors-To: ltp-list-bounces@lists.sourceforge.net To: Stanislav Kholmanskikh Cc: vasily.isaenko@oracle.com, ltp-list@lists.sourceforge.net Hi! > > On my systems /proc/sys/kernel/pid_max is set to 32768 and MSGMNI is > > from 2000 to 7000 on most of the machines I've seen. So the value of > > maxnkids was between 1 and 4 previously but it defaults to 10 after this > > patch which increases the test runtime by minute or two. > > > > Thinking of this again the most behavior preserving change would be: > > > > if ((MSGMNI * MAXNKIDS * 2) > (free_pids / 2)) > > maxnkids = MAX(1, ((free_pids / 4) / MSGMNI)); > > > > Which would scale the number of kids to the free pids but will not abort > > the test if there is not enough free pids. The drawback is that the code > > is complicated enough even without this change... > > > > According to the definition of recompute_msgmni() function (ipc/msg.c) > the limit by which the kernel may increase msgmni automatically is > IPCMNI. And IPCMNI is defined to be 32768 (include/linux/ipc.h). > > So yes, the above approach (using MAX) should work on most systems > (maybe even all. I'm not sure about embedded stuff) where neither > pid_max nor msgmni values were modified manually. Manually changing > pid_max and/or msgmni values we can break the test. > > Actually, I don't like the MAX-variant, because of the drawback you > mentioned. But the MAX-variant is sufficient for my needs. > > So which patch set to prepare? :) I would be inclined to set the default maxnkids value to something more reasonable than 10, perhaps 2 or 3 would be a good compromise between the test runtime and the amount of system stress the test does. -- Cyril Hrubis chrubis@suse.cz ------------------------------------------------------------------------------ Flow-based real-time traffic analytics software. Cisco certified tool. Monitor traffic, SLAs, QoS, Medianet, WAAS etc. with NetFlow Analyzer Customize your own dashboards, set traffic alerts and generate reports. Network behavioral analysis & security monitoring. All-in-one tool. http://pubads.g.doubleclick.net/gampad/clk?id=126839071&iu=/4140/ostg.clktrk _______________________________________________ Ltp-list mailing list Ltp-list@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/ltp-list