Messages in this thread | | | Date | Tue, 21 Sep 1999 04:48:59 -0700 | From | Brian Strand <> | Subject | Re: 1000ms delay in networking stack or driver, new bug |
| |
<delurk> I have observed this sporadically over the past few weeks, between my Linux and nt (I have to keep a couple around for my last remaining non-Linux-based project) boxen.
Linux box 1: RH6 kernel 2.2.5-15 (stock out of the box) kernel on a K6-2 300, upgraded from RH5.1, two interfaces: eth0, a Tulip-based netgear, eth1, 3c509-Boomerang. This box is running ipchains with eth1 managed by pump, and eth0 serving dhcp to the other machines. I upgraded this machine to kernel 2.2.12, as well as trying the latest drivers from Donald Becker's site (I'm relieved to hear about this from other people, since that means Donald &co's drivers are of the usual excellent quality (thanks Donald!)). I have seen problems with both interfaces being dog-slow but I had chalked up the 3c509's slowness to our (cable-modem-ish) ISP. On the tulip, I saw the same thing other people did: 1000ms pings, as well as "floods" of lost pings coming back all at once with response times very close to multiples of 1000 ms.
Linux box 2: RH6 PII-400 upgraded to 2.2.12, pump-managed 3c509C (Tornado?) with latest drivers.
The ping delay occurs when pinging both ways between Linux 1 and Linux 2 and from Linux 1 to either nt box. I haven't tried from nt to Linux yet.
/etc/rc.d/init.d/network restart generally clears things up. I haven't yet fired up tcpdump but will attempt it next time this happens.
Regards, Brian Strand </delurk>
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |