FD110 credit card terminal will not link up to an EXOS 5520 switch

[Keith9]

We have First Data FD110 credit card terminals that will not link up to an EXOS 5520 switch.  On a Cisco switch they auto negotiate to 100/full.  On an EXOS switch they also auto negotiate to 100/full but the device will not get a DHCP address.  The same machine works fine an an X450G2 switch from Extreme.

Whats really unusual is the switch shows a link, but either doing DHCP or Static IP on the credit card terminal, neither will ping out.  We are not sure what is so unique about these devices.  For now we have to uplink to a Cisco switch just to hold them.

Any idea what the issue could be?  

Thanks!

[Keith9]

GTAC Case # (02332940) has been opened for this issue.

Issue persists on two different 5520 stacks in two seperate geographic areas (ruling out hardware).

Both stacks are on 31.1.1.3.

Affected devices:
APC SmartUPS Rack mount UPS ethernet (requires static 10/full)
APC PDU addressable rack mount PDU's (requires static 10/full)
Audiocodes MP-118 VOIP Telephony gateway (requires static 10/full)
Barix extreamer100 audio streaming device (requires static 10/full)
FirstData 110 Credit Card terminal (requires static 10/half)

I was able to remove the Cisco 3750 switch today after finding this out.

Hopefully GTAC can resolve.  I do not have this issue on our X690’s or X450G2’s.  There is nothing special about these access ports. 

[yas1]

I was also very surprised that only static 10 Mbit worked with these devices. I hope that this is only a bad workaround.. Please keep this thread updated when GTAC provides some information about this.

[Keith9]

Here’s another strange one, so an APC rack mount SmartUPS was showing connected 100/full, but could not ping it.  I tried your trick to set it to 10/full and now I can ping it and access its web management page.

However on the switch its shows 10/full, 

but on APC’s web UI it says its 10/half.  

This is just so bizarre.  I requested EXOS 31.2.1.1 (upgrade from 31.1.1.3) so I will load that on later in the week and see if that changes anything.

[Keith9]

I recall not seeing a mac address at all on a port where a card terminal was installed.

However the APC gear like PDU’s and rack mount UPS’s do show.  Also I can verify the IP with sh iparp command.  There’s just no communication at all.  Cannot ping or access them at all despite showing 100 full.

Here’s an example of a APC Addressable PDU plugged into port 2:6 of a 5520 switch running EXOS 31.1.1.3 showing linked at 100/full

sh port 2:47 advertised

sh port 2:6 advertised
       10BaseT     100BaseT    100X        1000BaseT   1000X       Pause
Port   Half  Full  Half  Full  Half  Full  Half  Full  Half  Full  TX  RX  Asym
-----  ----------  ----------  ----------  ----------  ----------  ------------
*2:6   CLR   CLR   CLR   CLR   ---   ---   ---   CL-   ---   ---   CLR CLR CLR
Flags: (*) Active, (!) Disabled, (C) Capable, (L) Local Advertised, (R) Remote Advertised.

I’m also showing IP to ARP mapping

sh iparp | i 10.40.0.15
VR-Default    10.40.0.15       00:c0:b7:5a:e9:5f   20      NO  Default       1     2:6

sh fdb port 2:6
MAC                                      VLAN Name( Tag)  Age  Flags          Port / Virtual Port List
------------------------------------------------------------------------------------------------------
00:c0:b7:5a:e9:5f                          Default(0001) 0000  d mi           2:6

ping 10.40.0.15
Ping(ICMP) 10.40.0.15: 4 packets, 8 data bytes, interval 1 second(s).
Request timed out
Request timed out
Request timed out
Request timed out

--- 10.40.0.15 ping statistics ---
4 packets transmitted, 0 packets received, 100% loss
round-trip min/avg/max = 0/0/0 ms

So let me try forcing it to 10/full…

configure port 2:6 auto off speed 10 duplex full

ping 10.40.0.15
Ping(ICMP) 10.40.0.15: 4 packets, 8 data bytes, interval 1 second(s).
16 bytes from 10.40.0.15: icmp_seq=0 ttl=255 time=10 ms
16 bytes from 10.40.0.15: icmp_seq=1 ttl=255 time=4.003 ms
16 bytes from 10.40.0.15: icmp_seq=2 ttl=255 time=4.799 ms
16 bytes from 10.40.0.15: icmp_seq=3 ttl=255 time=6.238 ms

--- 10.40.0.15 ping statistics ---
4 packets transmitted, 4 packets received, 0% loss
round-trip min/avg/max = 4/6/10 ms

So now its connected.  This is so bizarre.  I will open a case with GTAC on Monday morning.  You shouldn’t have to run circles around managing port speeds.  We didn’t ever have to worry about port speed/duplex settings on Cisco 3750’s, Extreme X690s, or X450-G2’s.  The 5520 is supposed to be the newer, better switch.  To me this is a step backwards to the 90’s and earlier (having to manage speed and duplex settings).  Have to say I’m dissapointed in the 5520.  Between this and the inability to use 1000 base T copper SFP’s in the VIM-4X, I wish I would have just gotten the X450-G2’s.  They work as expected.

Thanks for your trick on specifying port speeds to 10/full.  At least that seems to get PDU’s and UPS’s on line.  When I’m in the office Monday I’ll move some of those troublesome devices off the Cisco switch and back to the 5520 and try the port speed there.

[yas1]

Hi Keith,

Do you see MAC addresses on these specific ports (sho fdb port <port>)?

Interestingly i had a similar issue on the new 5520 with a card terminal and a sensor which only works now with static 10/full duplex configuration. However on the old switch it worked with auto negotiation 100/Full.

Have a check with “show port <port> advertise” what the device is advertising to the switch and give it a try to set the port speed/duplex settings manually.

Best regards
Yannick