From: Carlos M. <car...@gm...> - 2005-10-23 15:58:17
|
On 23/10/05, Denis Vlasenko <vd...@il...> wrote: > On Sunday 23 October 2005 15:51, Carlos Martin wrote: > > On 23/10/05, Denis Vlasenko <vd...@il...> wrote: > > > if it doesn't do it. > > > > > > I guess it's kernel msg buffer overflow (klog may be too slow at read= ing > > > the ring buffer) > > > > It does sound like it. The loss happens when the rate goes up as well, > > at which point the messages would come in at a much higher speed. > > Try to bump up CONFIG_LOG_BUF_SHIFT=3Dnn in kernel .config, > and also you may turn off L_FUNC bit in acx debug mask > (will make log much shorter). The shift is already at 17. I'll take the L_FUNC bit out. > > > On a related note, my router is now using the ACX100 USB device. From > > tomorrow onwards, my Internet access will most likely be only through > > this device, so I won't be able to test it as often as I could before. > > If I'm lucky, I'll be able to use another card, but I wouldn't count > > on it. > > Whew, an acx usb in production environment :) It's working alright for now. > > Please find attached tarball. Since after looking at your logs > I am a bit puzzled I'd like to have requested rate to be recorded > into hostdata so that I san see it in struct txstatus. Now I have to find a time to do an scheduled downtime ;) which isn't going to be easy given my brother's downloading habits. I'll get some logs to you... eventually cmn -- Carlos Mart=EDn Nieto http://www.cmartin.tk "=BFHan entendido?" "S=ED, nosotros vemos La 2" -- Emilio, "Aqu=ED no hay quien viva" |