Also tried waiting .3 seconds after failure and attempting another SELECT
The Transaction instance has an ID, and the save method didn't raise an error, so why can't the ORM find it?
Scale Postgres to multiple nodes
Are the nodes updated instantaneously?
In this case, the INSERT appears to replicate to the SELECT node after the SELECT has already failed
Why do we need to SELECT immediately after the INSERT?
# refresh the transaction from the database since # a bunch of casting (in the process of insertion into the db) # probably occurred t = Transaction.objects.get(id=t.id)
What if we just stop doing that?
In general, minimize DB queries to reduce load on the DB
Specifically, try not to rapidly INSERT/SELECT, as the PGPool nodes won't replicate fast enough