Just a quick heads up for anyone using vDSwitches. I’ve ran into two issues and I would like to share it with those I’ve spoken with.
#1 – “load based” load balance/teaming policy.
#2 – vDS health check and physical switch Mac Address Table issues.
Here is the KB about the current bug fixes in a patch release (which will also be rolled up into 6.0 update 1).
Here is the text from the KB:
“When using load balancing based on physical NIC load on VDS 6.0, if one of the uplinks is disconnected or shut down, failover is not initiated.”
— This means, if you have a vDS, and “load based” teaming policy set on your 6.x ESXi host, then you remove a network adapter/Uplink, (or that link fails) the VMs will not failover or start using the other uplinks. This can and will cause an outage. The simple fix is to set the vDS teaming policy to the default “Route based on originating virtual port” or something other than Physical NIC load.
Just to clarify, this is not a vSphere 6 vDS issue, this is a host level – ESXi v6.0 issue. This can still happen if you have a 5.5 vDS and your host is running ESXi v6.
Here is the KB describing the issue.
The issue is when you enable vDS Health check and your vDS is large enough to over flow your mac address tables, up stream, on your physical network devices. This can, and will, cause an outage on the network. I have not tested or reviewed every switch mac address table limitations but anyone can reproduce this with enough effort.
So, how is this happening under the covers? When you enable vDS health check, it creates additional virtual mac addresses for each physical Network adapter attached to the vDS. It then sends out “packets” on all uplinks, on all hosts, on all vlans, and all port groups for that vDS. The text from the KB:
“There is some scaling limitation to the network health check. The distributed switch network health check generates one MAC address for each uplink on a distributed switch for each VLAN multiplied by the number of hosts in the distributed switch to be added to the upstream physical switch MAC table. For example, for a DVS having 2 uplinks, with 35 VLANs across 60 hosts, the calculation is 2 * 35 * 60 = 4200 MAC table entries on the upstream physical switch.”
So, lets scale that out further. If you have a 64 host cluster, each host has four uplinks attached to the vDS, all on a single vDS with 40 port groups. 64 X 4 X 40 = 10,240 mac address entries just slammed into your switch mac address table.
This might not be an issue for small businesses with small host and NIC counts but that really depends on the switch and router types they are using.
If you have any questions please reach out to me on twitter @vmnick0.me