The problem is s1 is down srv1 will not send the traffic
as long as they are different domains that is a normal setup.
as per the diagram, the server does not have dual-homed (its orphan port) - But the path is resilient and able to reach the server if your VLAN and Routing stretched..
You have to be more specific about the problem you have:
1. Can the server ping the gateway?
2. Can the server ping something in the same subnet/vlan?
3. Can the server ping something in different vlan?
4. Is it the only (single homed) server affected when S1 goes down?
5. What happens if you move the srv to sw2? Does the problem persists?
6. What's the gateway configured on the server and where and how it is configured? (example: is the GW the HSRP VIP?)
here is the test I carried out
I am not sure this is a coincidence or not. I shutdown the peer link , After that I removed both the members of vpc101.
I lost access to the loopback ip and also I noticed all the links connected from the N9K primary became orange including the links connected to access layer
This time i did not test the server access
In my case vlan does not have interface ip on the nexus switch .the ip's are configured on the firewall
Shutting down the peer-link is not the same as testing a reload of a peer. Also, removing a port-channel from vPC, it will enforce different forwarding mechanism compared with vpc enabled devices.
I would suggest you leave the vPC configured and up, and simply reload one by one the vpc peers.
Let us know of the results.