IE-3300 problem with 10M half duplex

[Steven Shelton]

We have six gas meters that have worked without issue for over eight years. Four were supported by IE-3000 switches.  Recently, we replaced two of the IE-3000 switches with IE-3300 and within hours began having problems with loss of connectivity to the meters.  When the meter fails, the mac address has aged out of the mac address table.  Resetting the switch interface restores connectivity until the next failure.  In one instance after resetting the interface, the meter became unreachable.  The meter's mac address was in the mac table.  A continuous ping for the meter was running,  the interface counters were cleared and shown no outbound packets on the interface facing the meter.  The mac table was checked again and the meter's mac address was still present.  The mac address eventually aged out.

As a test we reinstalled one of the IE-3000 switches to compare the interface errors with the IE-3300.  The IE-3000 switch was left in place for over a month with no problems. The IE-3000 meter does not log CRC erros but does log other errors associated with half duplex.  The IE-3300 logs CRC errors along with the other errors.

I understand the IE-3300 ASIC is unique to this platform.  Does anyone have information on the ASIC used in the IE-3300 and the IE-3000?  My suspicion is the IE-3300 is having issues calculating the Ethernet checksum and that's the source of the CRC errors.

Does anyone else have 10M half duplex devices supported by a IE-3300 switch?

I have an open TAC case for almost a month and TAC can't get past the layer 1 issues.  They also ignore the fact that the same cable and meter work without loss of connectivity with the IE-3000.

[jlpiccoli]

We have been struggling with 10mb/half duplex FCS errors and drops with ancient Allen Bradley PLC5 controllers. What corrected the issue for us was setting "no ptp enable" on the port. You can see if PTP protocol is present on the switch by doing show ptp (various commands). The default mode for PTP is e2transparent which will recognize a time sync master if present on the network and forward this packet data to interfaces unless specified not to do so. May be worth looking at.  

[Jeanmaia]

Hi, here we have a same problem with IE3300. My devices (load controllers) works in 10mb half and his interfaces in ie3300 have a lot of fcs errors. In cisco IE3000 this issue not happen.

We open a ticket on TAC and Cisco suggested apply "no ptp enable" in this interface  too. This workaround is working for us at moment. But we staying testing.

[Steven Shelton]

Thanks for the input.  On 4-11-2023, jlpiccoli, made the same suggestion and my port configuration did not have no ptp enable.  I disabled ptp on the ports facing the 10M devices.  We continue to be plagued with this issue and continue with port resets to reestablish connectivity with 10M devices.  I had to reset a port this morning.  Good luck, hope it works for you.  Cisco product development obviously did not expect 10M devices to exist and failed to test the ASIC properly.