Unable to ping other subnet IP, until clear ip arp issued in CoreSwC1k

[aditya@network-ie.com]

Hi All,

We have Cisco C1000 Switch (IOS 15.2(7)E5) working as Core in our LAN infra. We have multiple vlans configured, interface vlans are configured on the same C1000 Switch.

We are facing strange issue, intermittently hosts connected to access network are unable to ping the other subnets apart from their own vlan (subnet) able to ping gateway as well.

When we check the show ip arp and show mac address table it shows proper mac entry and ip arp entry.

Once we issue "clear ip arp XXXX.XXXX.XXXX". it starts pinging in other subnets as well.

Or

If we add a static entry for ip and MAC, that also makes pinging in other subnets to work.

Kindly assist to find the issue and resolution. 

Regards,

Aditya Deshmukh

[Reza Sharifi]

Hi,

Anything in the switch logs that indicates a problem?

HTH 

[aditya@network-ie.com]

Hi,

Nothing all logs are clean.

[mlund]

If someone has configured a pc with the same ip as the gateway. Then when pinging the local subnet all is working. Also when ping the gateway works, because the rough pc answers. But when ping outside, the ping goes to the rough pc instead of the gateway. Look at the arp in the client and see if the gateway arp is changed when traffic interrupts.

[MHM Cisco World]

are you config DHCP snooping with DAI ??

[Metzge]

We have very similar issue.  I believe it's a bug and should be resolved by going to 15.2(7)E7.  Testing that now.

Did you every solve the issue?

[Metzge]

Follow up to our issue.  One other thing to consider is the SDM profile.  It's really small for the C1000.  I believe it's 512 total.  So that means the sum of active hosts on your vlans can't be over 512 to be able to route to them.  The 2960XR I believe was at least 4K.  Another thing is it doesn't log anything either when it hits this limit.  You just get random hosts that can't route off the local vlan they're  on.