Obligatory "I am new to configuring Cisco switches" comment and I apologise if I am asking a silly question.
The issue in a nutshell is that when configuring a stack of two Catalyst 3850 switches, if I convert a trunk port with one VLAN allowed to an access port with the same VLAN I can no longer ping it.
Here is the environment.
Server A is a Windows 2012 server with NICs connected to ports g1/0/1 and g2/0/1 of the switch stack. An etherchannel is configured, the ports are in mode active and in trunk mode with VLANs 101 and 102 allowed. On the server side the NICs are teamed and VLAN tagging is enabled for two interfaces, management (101) and production (102). The management interface has an address of 192.168.0.1. Everything seems fine with this configuration. The etherchannel is up with both interfaces in status P.
Server B is an ESX server. The management interface (VLAN 101) has two NICs connected to ports g1/0/2 and g2/0/2 of the same switch stack. The NICs are not teamed. IP address of the interface is 192.168.0.2. On the switch no etherchannel is configured, the ports are in mode trunk allowed VLAN 101.
With the above configuration I can ping from server A to server B. But if I set the ports g1/0/2 and g2/0/2 to mode access VLAN 101 I cannot ping. Why not?
Also what is the downside (if any) to have the ports in trunk mode instead of access mode?
Whilst troubleshooting I have shutdown port g2/0/2 to simplify the configuration but with the same result, server A cannot ping server B unless
