Is it possible to put ping6 on the second interface of one of the machines connected to the network?

Question

linux

There are two vm (linux) connected in an IPv6 network.

vm1:
eth0: 2000::1/3
eth1: 2000::3/3

​​vm2:
eth0: 2000::2/3

Looks like this:
vm1:eth0 - host:br0(br0-dummy) - vm2:eth0

Virtual machines see each other (2000::1 and 2000::2) mutually exchange ICMP. When trying to send an ICMP packet from vm2 to vm1 requesting address 2000::3, the packet with multicast ff02::1:ff00:3 stops on interface eth0.

With IPv4, a similar scheme works.

Answer 1

Stops at eth0 VM1? Then everything is fine.
What do you want? The packet reaches the desired host, there is an answer.
Or do you want the sender address of the response to contain the address to which the packet was sent? But this is not the case with IP4 everywhere. Stack implementation specifics. The response is not generated by the network adapter, but by the OS.
PS: if I'm not mistaken, in Windows XP it was impossible to ping the second interface at all - it simply did not respond to such requests. As in modern versions - I do not know.

Answer 2

what do you need from ping ?
ask stupidly whether there is availability ?
ICMP and broadcasting are not the best helper for working with interfaces, even for the sole reason that tcp \ ip is everywhere, which is an order of magnitude higher.
If we take any utility already working in the stack.
Then you will easily get the same result. hping3 telnet
example and a million other things that could give a response time or availability. Maybe you shouldn't bother with ping?