sam_builds Importing a 20,000-row CSV through an Edge Function. It dies at 150 seconds with about 8,000 rows done.
Increased the batch size to 1,000 rows per insert. Slightly faster, but it still times out around row 8,000.
Edge Functions have a hard wall-clock limit; you cannot raise it. Make the work fit:
copy via a pg_net trigger or supabase db CLI. Postgres eats 20k rows in under a second.For a one-off import, just run psql \copy from your laptop.
Small correction on the first option: Postgres can't copy a file that sits in Storage. COPY FROM reads files on the database server, and your CSV isn't there. So for a recurring import, go with the split-and-resume approach, or stream the file from Storage in a function and insert in batches of a few thousand.
For a one-off, \copy from your laptop is the answer. It sends the file over your connection, which is the whole trick.
sam_builds
\copyfrom my laptop: 20,114 rows in about 3 seconds. I'd spent a day on the Edge Function.