@al2o3cr Thanks for the response!
Not actually 10s of millions but just estimate 40,000 rows from all tables. You are right about two things:
- There are transactions that are taking too long thus result to DB timeout
- If insertions will be through GenServer processes and DB is the bottleneck, the process will just wait for DB to open a connection
We need to update our seed from time to time when new features are introduced and sometimes when we update a function in the seed it can result to timeout which occurs intermittently. When we (especially QA) want a fresh seed to our database, and it times out, we need to fix the seed and re-run it from the beginning.
So I think it will be better to just run the seeds sequentially to avoid overloading the database? Also it will be easier to track where it stopped and where to fix
