oh cool great my dataset failed to import at 31 million out of 38 million matching candidates, after searching through 8TB and indexing 250GB of metainfo into postgres, and now i can basically start anew - not because it would be hard to recover the lost progress but because i entirely overengineered the shit out of this deployment and it automatically scales up to hundreds of 16 core 32gb ram cloud servers because i got tired of being ratelimited when writing shell scripts to wrap around opensearch-cli so i wrote a pure rust+nix+nushell "deployment controller" to speed things up lmao