Hello All,
I apologize if this has been posted before and answered elsewhere, I've searched and was unable to find anything discussing the same situation.
I have a Windows NLB setup in our environment on 4 hosts in a blade chassis. I have 2 port channel links going to the blade chassis from my core to the left/right side of the blade chassis with multiple cables. I've followed the documentation for creating the static ARP & MAC entries on my core switch so that I can ping the VIP on the NLB.
All was good for a couple of weeks, then I started experiencing intermittent connectivity loss, eventually 100% connectivity loss. I started digging in to the environment, and when the static entries were created, they were configured on port channel 1, which is where both servers real ARP/MAC/IP addresses were living on at that time. However, now it appears that the ARP/MAC/IP addresses for the server real IPs are coming down on port channel 2. I've since switched the entry on the core switch to point the VIP ARP/MAC to port channel 2, and I was able to connect without issue again.
So, how can I implement a permanent fix for this issue?
Do I need to add the ARP/MAC addresses to one of the blade switches itself? I think this could lead to the same problem in the future if it's added to 1, and the servers are attempting to use 2.
Do I need to do something on VMware to force all connectivity for those 2 real server IPs down just port channel 1? Would I need to re-configure my entire networking setup for VMware hosts, new virtual network switch, etc ?
Thanks in advance for any support you can provide!
I found out that my switches do not allow me to add the mac address entry on both port channels. As a work around, i created a new port group with an active / standby nic instead of active / active. This will force all traffic down port channel 1 unless that entire port channel goes down. This is my work around until my switches support multiple mac addresses on both port channels.
No further responses required on this.
Thanks