I just upgraded my Postgres DB on Heroku and used pg:copy to copy everything. However in my original DB it shows there was ~32,000 rows and in the new one there are ~28,000 rows. Does this mean some rows weren't copied or is this common ?
The variance in row counts can be attributed to stale data, table bloat, or both. pg:copy uses pg_dump and pg_restore under the hood. If pg:copy completed without errors you can rest assured that data was not lost in transit.