Windows failover clusters and IGMP snooping

This post was originally published on this site

I’m trying to get Windows Failover Clustering up and going in our vSphere environment, and it’s driving me nuts.  I can successfully install the cluster after passing the Cluster Verification Wizard, but I keep on getting cluster failures where the VM heartbeats are getting dropped.  I look at the cluster logs, and see errors like  “LocalEndpoint blahblah:~3343~ has missed two-fifth consecutive heartbeats from yakyak:~3343~” which seems to be a network issue, without the heartbeats being totally dropped.  I’ve gotten to the point where I just have two Server 2016 VMs running with failover clustering installed, two NICs with a client and a cluster interface, a File Share Witness on an SMB share, and nothing else.  No roles, no shared disks, no agents, and I put them on the same ESXi host.  This is why I’m posting here, since the traffic should not be leaving the host at this point.  Some background: we’re running vCenter 6.7, and these VMs are on HW version 15.  After some research, I’m wondering if I need to turn on IGMP snooping for these heartbeats to be consistent?  This would also have to be enabled on a vDS that has like a hundred other VMs; would that possibly cause issues with these other VMs?  I’m just trying to understand what I’m missing here.  TIA

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.