I have noticed that if I get disconnected at the keepalive time (/proc/sys/net/ipv4/tcp_keepalive_time) -- I wound it down to about 10 minutes and got disconnected at the keepalive time on the dot.
jch
Christopher Wong wrote:
Here's some puzzling behavior. I have 3 machines running Red Hat Linux on our LAN that spontaneously and simultaneously get "disconnected" from the network. That is, they remain responsive and seem to hum along fine, but they cannot see any incoming network traffic. I ran ifconfig repeatedly and saw that the "dropped" RX packet count increases steadily (actual RX packets do not change). Shutting down iptables has no effect. "service network restart" will get eth0 going again, at least until the next failure.
Any ideas as to the cause? What these machines (RH9, RH9 and RH7.2) have in common is that they are running 2.4.20 kernels (2.4.20-9 and 2.4.20-13.7) and ntpd. Other PCs running a variety of OS and kernels are unaffected. Suggestions for investigation? Thanks.
Chris
