Monitoring a suspect NIC / workstation

We recently had a system wide network issue where we suspected one machine was causing the problem because it was the last thing changed on the network and when we unplugged it the problems sunsided.

We also reset the switch it was connected to which was showing down in SW at the time. I want to put that PC back on the LAN but want to setup a monitor so that I can be warned if it starts misbehaving

What tye of monitor whcould we configure? I see in NPM I have options for CPU,Memeory, etc. along bandwidth utilization, packet loss, latency, errors, discards.

What parameters should i set for discards, errors, etc to trigger an alert that this NIC is going to cause issues?

Thanks
Mike

Find more posts tagged with

troubleshooting_network_events

perfornamce

troubleshooting

nics

Accepted answers

All comments

mharvey

Mike,

The errors and discards may not prove to be as helpful as you think. This are measurements of bad and/or dropped packets in the inbound and outbound traffic on the NIC. What signs was that PC having when there were issues? Was there high CPU? Bandwidth from the NIC? You'd need to establish that before deciding what to monitor. I'd go ahead and add the workstation and monitor everything for traffic, cpu, and memory, and keep an eye on that machine. What changes were made on that machine prior to the network having issues? This would better determine what to monitor, or at least, what to focus on and setup an alert for.

Regards,

Matthew Harvey

Loop1 Systems

http://www.loop1systems.com

mhamill

Actually in the midst of a severe network outage where we could not see what the issue was, we asked ourselves - What changed recently? The answer was the addition of this machine the previous day.

We then noticed via SW that the switch that this unit was connected to was showing as down. We unplugged the new unit and bounced the switch at which point all of the network issues were resolved.

We need to add that PC back tot he network and wanted to monitor for trouble in advance of any cumulative problem which might again take the switch out.

mharvey

Then I would start with just CPU, Memory, and Traffic and keep an eye on them to see if anything starts becoming an issue when the machine is added back.