What causes high network error rate?

[Mladen Despic]

Lately, we've been receiving Measureware alarms from
a system (running HP-UX 10.20) on a daily basis:

"network error rate is greater than 10 per minute"

In fact, we can observe, in PerfView, that GBL_NET_ERROR_1_MIN_RATE is as high as 300 at times. For most of the time, however, this metric is at 0.

I am attaching the output from 'lanadmin' for the primary network interface. Obviously some errors have been logged, but I don't understand what they mean.

Any clues or suggestions? Thanks.

Mladen

[Mladen Despic]

# netstat -in
Name Mtu Network Address Ipkts Ierrs Opkts Oerrs Coll
ni0* 0 none none 0 0 0 0 0
ni1* 0 none none 0 0 0 0 0
lo0 4608 127 127.0.0.1 5281849 0 5281849 0 0
lan0 4352 165.23.4.0 165.23.4.10 240685 0 218729 0 0
lan2* 1500 none none 0 0 0 0 0
lan1 1500 165.23.5.0 165.23.5.10 156955170 16262 280673594 4 0

[Mark Bowden]

Since your attachment only shows errors on the input of the interface, I would suspect that some other device on your network is generating the traffic that is causing the alarms.

[Patrick Wallek]

One thing to check just to make sure it is not the problem is that the switch or router port that this machine is plugged into is set to 100 Full Duplex. A duplex mis-match between the machine and the switch / router can cause some strange problems.

[Glen Van Lehn]

Mladin,
do not overlook the cables and switch port.
you say, "Lately," implying that the system was running normally without errors in 'recent' past. Full/half duplex problems ARE a pain when 'auto-negotiate' doesn't, but usually you see collisions as well as Rx errors, but do that first. Then, i'd check the switch port counters. You could see Rx errors at the switch which might mean a degenerated cable, or not, which could be a bad pair.

with structured wiring you often have two patch cords to deal with... and one just 'ajar' can create problems like this also.