From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andy Furniss Date: Fri, 26 Nov 2004 22:28:17 +0000 Subject: Re: [LARTC] HTB Script Message-Id: <41A7AE01.6020206@dsl.pipex.com> List-Id: References: <41A5065C.8060304@wp.pl> In-Reply-To: <41A5065C.8060304@wp.pl> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: lartc@vger.kernel.org Lenthir wrote: > Hi everybody! > > I wrote nice script, but I need some help.... > Script is working almost well. > > My question is: > Can I minimalize PING time more than this script can? (nowadays is > 70-150ms with large load of link) > My first problem is - unreal big PING on router, and almost excellent > (nowadays is 70-150ms) PING on computers in network. > > My users don't complain, but I noticed that the pages is becoming load > quickly and when it's almost done.. the transfer is going down!! > And second symptom is: when I open pages PING grows to 800ms, and > quickly go down. It's almost imperceptible, but it happens. > What's wrong? All packets is going to correct pipes. What should I do more? > I suspect, that time of change of speed (HTB's reaction time) many > queues is long. How can I improve this? It's a big script - I only looked briefly and may have missed things. Shaping inbound traffic from the wrong end of the bottleneck is hard, HTB would need to be predictive to do it better. For SFQ you can change queue length (SFQ_DEPTH) in net/sched/sch_sfq.c or you could use esfq and choose length with options. How many users do you have? If you have 2mbit link then I think you need to reduce the ceils or you won't have a queue to shape with. > Script is very nice, but I need some professional help and advices. > > Everything You can check when script is working: > ./htb0.5.3en stat (example: ./htb0.5.3en stat 192.168.0.2) > ./htb0.5.3en stat lan > ./htb0.5.3en stats > > My network's statistics: > http://stats.opat.hopto.org/ > http://stats.zabierzow.net/ > > I attached my script. IMQ with imq_nat.diff patch is required (or AB > option in kernels 2.6), patched iptables(IMQ patch) is required, iproute > is required and bc is required. I tested this script on 2.6.9 kernel and > iproute from .deb package. > > And second less important problem... How Can I mark squid's MISS > packets? Is Debian's Squid Package patched to change TOS? > http://stats.zabierzow.net/squid.php :d I assume a miss is traffic squid gets from the net, if unshaped this will make you go overlimits - so is not neccessarily just a secondary problem. If you are NATing and have IMQ hooking after NAT in PREROUTING then traffic headed for squid will still have your real IP and forwarded local IP. You need to use u32 to seperate it. Andy. > Someone will help? _______________________________________________ LARTC mailing list / LARTC@mailman.ds9a.nl http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/