Looking for a pointer or a direct explanation as to how chassis clustering works from the point of view of other devices on the network. That is, how does do peers on the Layer 2 link and the switch learn that traffic to the reth0 address now goes somewhere else?
Is it like VRRP where the other Layer 2 devices never see anything and the switch just needs to learn that the virtual MAC is somewhere else now? Does the MAC-IP pair change for the LAN neighbors and the switch really doesn't see anything?
If you are talking specifically about RETH, then yes, it is similar to VRRP where the SRX sends out a gratituitous ARP during a failover scenario to "take over" the IP address from the failed device. Chassis cluster also describes many other failover scenarios (l-3 for example with a routing protocol), as well as the general mechanism for control-plane failover.