Folks:
On our ESXi 6.7U3 hosts, the management Vswitch has 2 vmnic's for "redundancy". Each vmnic connects to a separate switch (server switch A, server switch B), these switches in turn are uplinked to our core switch. Our network team noticed that when the uplink from the one of the server switches to the core switch was disabled, we lost connectivity to several ESXi host servers.
Investigating further showed the active vmnic on the management vswitch was connected to the switch with the disabled uplink to the core switch. The network team asked what factors force vmnic's to be flagged as down. Or how to we implement a truly redundant connection to the ESXi management vswitch?
Mark M.
I assume that both vmnics are either configured as active/active, or active/standby on the vSwitch/port group level.
By default ESXi only checks the NIC's uplink state, so it won't detect failures of further connected switches as in your case. In order for ESXi to detect such failures you'd need to enable Beacon Probing, which however requires at least 3 uplinks.
An option in your case - if the switches support this - is to configure Link State Tracking, so that the port to the ESXi host is being shut down in case the uplink to the core switch goes down
André
Are both physical nics from the ESXi configured as active?
I assume that both vmnics are either configured as active/active, or active/standby on the vSwitch/port group level.
By default ESXi only checks the NIC's uplink state, so it won't detect failures of further connected switches as in your case. In order for ESXi to detect such failures you'd need to enable Beacon Probing, which however requires at least 3 uplinks.
An option in your case - if the switches support this - is to configure Link State Tracking, so that the port to the ESXi host is being shut down in case the uplink to the core switch goes down
André
Yes they are, I see the "network failure detection" is set to link status, now I see there is a 2dn option for Beacon Probing, looking into this option
I'll let the network team check into the Link State Tracking, since we don't have extra vmnics to add to the vswitch to attempt Beacon probing, thanks for the tip on this!