Glad you managed to solve it! I didn't notice your reply before my last answer, but yeah, I'll keep it so it's easy to fetch that command in the future hehe
Thanks for sharing the fix.
You can attempt running the below command [0] on all cluster nodes:
~# pvecm updatecerts
This should recreate the node specific ssh_known_hosts file based on its SSH key.
See man pvecm [0] for more details.
[0]...
Aha this make sense, Proxmox ve uses pmxcfs, its FUSE-based filesystem that stores configuration in a SQLite database and synchronizes it across the cluster.
There's a pve command to correct this, I will look for it and update you
What is the destination you are trying to reach? Is it the IP of a Virtual Machine, or an IP you assigned directly to the vmbr2 bridge? Also, from where are you testing the ping, source and destination
Yes you can start with bridge-vids 2-4094 as on the example. Later if this help to bring the interface working, you can adjusted to allow only the needed VLAN, as this advised from both security and performance wise.
Regarding the second...
I don't have experience with Cisco UCS and VIC, but I have came across some thread [0] mentioning that marking the bridge devices as VLAN aware [1] will make the VIC cards work as designed. So I think you can attempt adding...
Yes, the answer is Yes.
Its needed for trust, because other wise Man in middle attack could occurs. It's also needed for compliance and regulations. But indeed if your environment is isolated then self signed or even expired certificate might be okay