-
Bug
-
Resolution: Unresolved
-
Normal
-
None
-
4.14.z
-
Moderate
-
None
-
AI-51
-
1
-
False
-
-
Customer Escalated
Description of problem:
In Agent-Base Installation, storage network on all nodes will be also configured. If both VLAN interfaces in the same L3 switch used as gateways for compute cluster management and storage networks of the OCP cluster have arp-proxy enabled, then the IP collisions validation will report errors. The reason why the IP collision validation fails is that the validation seems to send the arp-request from both interfaces bond0.4082 and bond1.2716 for all addresses used for compute cluster management and storage networks.
Version-Release number of selected component (if applicable):
4.14
How reproducible:
Always
Steps to Reproduce:
1. Install cluster using Agent-base with proxy enabled. 2. Configur both VLAN interfaces in the same L3 switch, used as gateways for compute cluster management and storage networks of the OCP cluster have arp-proxy enabled
Actual results:
IP collision validation fails, the validation seems to send the arp-request from both interfaces bond0.4082 and bond1.2716 for all addresses used for compute cluster management and storage networks. The validation seems to trigger the arp-requests sent out from all NICs of the nodes. If the gateways connected to the different NICs of the node have arp-proxy configured, then the IP collision failure will be observed.
Expected results:
Validation should pass in Arp Porxy Scenario
Additional info:
In arp-proxy scenario, the arp reply is not needed from the NIC which IP address is not in the same subnet as the destination IP address of the arp-request. Generally speaking, when the node tries to communicate to any destination IP address in the same subnet as the IP address of one NIC, it will only send out the arp-request from this NIC. So even arp-proxy in this case, it will cause any issue.