not an expert here. My approach here would be to at least move this to the supervision tree by using Task module (Task hexdocs) and appending it right after the registration of DB process.
Extra step could be to make another Supervisor, that would wrap/manage both DB and this one-time seed Task processes, so that (as mentioned by NobbZ) when the DB fails, the new DB respawns and the intermediate DB-supervisor (with strategy: :rest_for_one Supervisor strategies) could start the seed Task after DB successfully respawns.
But that might not be, how it should be done.. Just saying, how I would approach this and leverage OTP supervision trees.
