Maybe you can get through this without changing your hardware, by enqueing the outgoing packets on the linux box, not on the modem. This can be done using QOS - cbq or tbf (which have to be compiled in the kernel or as modules) to slow down and enqueue the packets going out through the ppp. You also need iproute2(tc is the program that you really need) to configure the packet scheduler's behaviour. Documentation can be found at http://lartc.org or http://vasile.lantech.ro/pub/Manuale/lartc.pdf (the last one is saved on my computer at home but is available 24/7 and you'll find some good examples there - see section 9.2.2.2, 9.2.3.2 - you should try this first and more examples if you really want to learn bout the power embedded in the linux kernel ;)) )
Statu Nascendi Master of Disaster ----- Original Message ----- From: "Russell Coker" <[EMAIL PROTECTED]> To: "Simon Bland" <[EMAIL PROTECTED]>; <debian-isp@lists.debian.org> Sent: Wednesday, October 02, 2002 10:44 AM Subject: Re: Odd network behaviour > On Wed, 2 Oct 2002 05:02, Simon Bland wrote: > > I've recently upgrade my desktop and I've noticed some odd behaviour > > with my network since then.. My machine is currently acting as the > > gateway for a small home LAN, I'm running dnsmasq to handle the > > forwarding and iptables for the firewall. > > > > The odd behaviour is that any time my machine is under load the network > > cuts out all together, or only about 1 in 50 or so packets gets back > > from a ping. > > > > I should probably clarify that a little, the LAN side seems to work > > fine, it's the PPP interface thats locking up when I'm under load, which > > is sorta strange being that the modem is about the only thing I didn't > > change. > > I used to have the same problem on a 386-25 with a 19K2 modem (38K4 port > speed) on a serial port without a 16550 buffered UART. When the machine was > not under load it could respond fast enough to interrupts, when the hard > drive was busy it missed characters and all the data it received was > scrambled. > > This is a symptom of hardware that sucks. > > The pre-emptable kernel patch may alleviate this. But improving the hardware > is the only real solution. > > -- > http://www.coker.com.au/selinux/ My NSA Security Enhanced Linux packages > http://www.coker.com.au/bonnie++/ Bonnie++ hard drive benchmark > http://www.coker.com.au/postal/ Postal SMTP/POP benchmark > http://www.coker.com.au/~russell/ My home page > > > -- > To UNSUBSCRIBE, email to [EMAIL PROTECTED] > with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED] >
