Well, dunno how a shared hub could actually be traced down other than seeing WHAT was bad and you divined them to be all plugged into it...

Well you coulda looked at the "uplink" you had and saw the utilization coming from it, the errors, the bursting etc.. going on, on that port. But iffin you ain't got those reglar nummers and infer bout wha they suppose to be.. yah ain't gonna know!

Recently I have been having tremendous supervisor module (for my VLAN router HAH!) LAG or it just peaking out it's processing capacity. Well as it turns out, we (still) have multiple issues with this particular brand of "router" layer-3 switching acting as a router doesn't really work as well as it could. This brand we were "forced" to purchase, the company was spun off into a telephone/networking/B*LL LABS company. Then the telephone/networking got spun off from the B*LL LABS. And it is currently not doing well in the "liquid assets" topic.

Well needless to say, the dang thing can't handle RIP routing any better than, oh say... A tele-typist.

Reason I say, default routes are default for a reason, you want to traffic to at least hit the route for a reason. Now, if you have 2 gateways on a Subnet... the lowest percieved route wins (as I put it, least cost RIP routing). Well that requires an ICMP redirect from the default route to the cheaper route. The main VR cannot handle it properly and never informs the originating TCP stack to remember the route for a specific time, it just tells it to go there... not me. So the processor peaks out trying to redirect thousand of connections from a squid/http/log/time server. Rest of the network falls on it's face...

We sniff the port, no errors other than slow response times (DUH). we use the switch "cool tools" available to use (not good ones that actually work the way they are supposed to) they tells us nothing... nada... zip... zilch... I switch the default route to point to the "firewall private interface" and all problems go away. Yet the VR is still handling the same amount of data, just not handling redirects. The *NIX based TCP stack works like a charm and I bitch about the VRR product we have, I get told to basically shut-up.

Nuff bout my stuff, as for yours.. use the snmp tools you got from your vendor or build some of your own (providing you have some manageable switches/hubs).

Other than "pinging" nodes on the various networking devices you have not too much out there.