I observed that my node EveLink has a single upstream connection to AS1 (10.0.5.1/30) and a loopback at 91.214.0.1/32. AS1 advertised 7 prefixes total across two batches: its own loopback, Uni (two addresses), AS2, and ACM (three prefixes). All destinations were reachable with 0% packet loss, with latency increasing linearly from ~20ms (AS1) to ~90ms (ACM).

I decided to install a default route via AS1 with `src 91.214.0.1`, plus specific /32 routes for each advertised prefix, and to source all pings from my loopback. I installed the second-batch prefix (128.173.10.1) promptly upon receipt.

The default route with loopback source ensures all egress traffic uses an address remote nodes can route back to, since link addresses aren't advertised network-wide. Specific routes provide explicit precedence and table clarity. I trusted the additional prefix from AS1 because it's my sole transit provider and a single /32 is consistent with a campus user address. I verified after every installation to confirm fixes directly rather than assume they work.