IBM AIX PowerHA problems with CISCO ACI

Hello

we found a serious issue about PowerHA 7.x under AIX 7.x . We migrated from CISCO Nexus to CISCO ACI (Application Centric Infrastructure - https://www.sdxcentral.com/data-center/definitions/what-is-cisco-aci/)

When you have a ressource group failover it can happen that the ip adresses on the new target are not reachable anymore . This means at the end of the day your application is not reachable anymore. This happens e.g. when on the Resource group source server an application like SSH etc is still sending packets to the LAN switch where the machine is connected to. Under Nexus PowerHA makes a GARP request - Gratuitous ARP and ignores the coming packets simply. Under ACI we have a combination of Mac Adress and IP Learning - See also CISCO Whitepaper ACI Fabric Endpoint Learning White Paper

In our case we have seen packet traffic on the PowerHA new Resource group target server and "old" resource group source server. This leads to a not reachable IP address on the new target server hosting now the application. For this we testing an IBM Fix to solve this ACI problem with two new options for AIX :

no -p -o ip_ifdelete_notify=1 and no -p -o ip_ifdelete_no_retrans=1

It seems that this fix will work for ACI with PowerHA

This information is on my own and using this information is on your own risc and has nothing to do with my company i'm working for.