Switch on Network and Looped to Itself Crashes Client Access

Monday at 4:15 p.m. users lost their "client access" sessions and could not re-establish connectivity. PC's couldn't ping the iSeries. Twinax connectivity was fine. I was able to get into the HMC from home and got a terminal session and surprisingly, there were no outstanding QSYSOPR messages. Our VP was able to telnet from our second partition into the one where client access was not working. IBM support said it was a network problem. Problem was limited to Client Access only. Our Novell file network was fine along with the VPN I used to connect to the HMC from home.

At 11 p.m., we finally solved this problem when we removed a 2 foot patch cable that was looped from port 1 to port 2 on the same $25 Cisco OfficeConnect Switch on the production floor. This switch connects a printer and a thin client to an HP ProCurve switch which connects to the CISCO switch that our iSeries interface connects to. A production worker was having trouble with the printer and tried to "help" by plugging the loose end of a patch cable into the switch.

Our network people think the iSeries just couldn't keep up with the traffic and that is why we couldn't use Client Access. But wouldn't there be some type of message indicating a threshold, or something of that nature, has been reached on the iSeries interface?

Bryan Burns
IBM Certified Specialist - iSeries System Command Operations V5R2