ARP announcement does not refresh the ARP on core switch

[seanxiao]

hi everyone,

our production line use wired scanner with static IP address, let's say scanner A is broken or reached EOL, and got replaced with scanner B, IP address will stay the same. But when the scanner B is connected with the same IP to the access switch on the same port, it cannot work. I cannot ping it from the core switch,  and if I show the arp by the IP address, it is showing the IP is mapped with Scanner A's mac address still. Then I have to manually clear the arp cache on core switch, only by which scanner B can works right away.

And then if I reconnect scanner A back to the port with the same IP, the same issue occurred to it. But this issue is not happening to laptops.

I do see scanner B and scanner A  sends ARP announcement after it beeps upon being connected.

pic 1: HP laptop sends arp announcement

pic 3 : Scanner A sends arp announcement:

pic 2: Scanner B sends arp announcement:

but if I connect a laptop with the same IP address and connect it to the same port, I can ping laptop right away from the core switch.

should I monitor the traffic on the port-channel that connects the access switch and core switch to see if the ARP announcement is passing through the port-channel? 

access switch is C9300 and it has been rebooted for troubleshooting.

Cisco IOS Software [Bengaluru], Catalyst L3 Switch Software (CAT9K_IOSXE), Version 17.6.5, RELEASE SOFTWARE (fc2)

thanks for help

[alancelliott]

Hi Sean,

Are you able to show us port configs, as well as anything that would effect this i.e. dhcp snooping, arp inspection? If you run debug arp on your core switch are you seeing the GARP ingress?

[MHM Cisco World]

It issue of scanner not send GARP and update mac in all l2 SW.

Connect PC in same subnet of scanner ping the broadcast of subnet and make scanner reply' this case the SW will update MAC 

Broadcast of subnet  is like below 

10.0.0.0/24 it broadcast is 10.0.0.255

M

[seanxiao]

hi all,

just an update, configuring spanning-tree portfast  on the access port will "fix" the issue, this is really insane. How spanning-tree become a must in network, especially only for this cognex scanner? What if a network environment without any STP? this is ridiculous.

[Ruben Cocheno]

@seanxiao 

That scanner is likely to be a silent type host, having snmp/syslog enable on that scanner will keep it alive on the network. But perhaps check the settings and double-check if something is blocking the GARP (e.g)

[seanxiao]

thanks Ruben, I found taht configuring spanning-tree portfast  on the access port will "fix" the issue, this is really insane. How spanning-tree become a must in network, especially only for this cognex scanner? What if a network environment without any STP? this is ridiculous.