Hi,
in the meantime I discovered, that setting one member-port of the VLAN to loopback like suggested above was a very bad idea:
The mac addresses from PCs connected to other VLAN member port start flapping between their connected port and the port set to loopback. This causes heavy connection interruptions of those PCs.
If you use "show ethernet-switching mac-learning-log" you'll see the flapping:
...:11:03 2011 vlan_name VLAN-3-LAN mac <mac-address@port5> was deleted on fe-0/0/2.0
...:11:03 2011 vlan_name VLAN-3-LAN mac <mac-address@port5> was learned on fe-0/0/5.0
...:11:04 2011 vlan_name VLAN-3-LAN mac <mac-address@port5> was deleted on fe-0/0/5.0
...:11:04 2011 vlan_name VLAN-3-LAN mac <mac-address@port5> was learned on fe-0/0/2.0
...:11:05 2011 vlan_name VLAN-3-LAN mac <mac-address@port5> was deleted on fe-0/0/2.0
...:11:05 2011 vlan_name VLAN-3-LAN mac <mac-address@port5> was learned on fe-0/0/5.0
fe-0/0/5.0 was the port a PC is connected and fe-0/0/2.0 was the port set to loopback.
So, using a loopback interface seems not to be a method to keep a VLAN always up.
Any other suggestions?
On some IOS switches Cisco uses "n keepalive", but I couldn't find a possibility to something similar on an SRX's VLAN interface.
- Steffen