Cd into setup_scripts and run sudo ./setup_[DBMS].sh to reset state of databases (REQUIRED ONCE PER SERVER-- to group members, you don't need to do this again for postgres and tpcc, I've already ran it). See example.py for usage examples.
"Template" databases are used to expedite reiniting the tpcc "clean" database (i.e. no tpcc additions or deletions). However, this means that we will have 4 copies of different databases in use at any given time, each ~2GB in size at scale factor 20. Additionally, this server has a disk size of around 8GB, so it is impossible to load all 4 copies in at the same time. As such, you will need to run a setup script, train + evaluate the RL-derivdd parameters, and then DROP the no longer needed databases as shown below. This applies to copies of tpch as well.
Run drop_everything.sh in setup_scripts-- BE VERY CAREFUL ABOUT THIS, YOU COULD BE UNDOING ALMOST AN HOUR OF WORK IF DONE ACCIDENTALLY.
Run setup_[pairing].sh (this will take a while. Tpcc ~5 min, tpch ~40 min)
Steps to reproduce in case the server gets nuked or recreating from scratch:
Clone and build https://github.com/electrum/tpch-dbgen (or steal the one in /home/rz97/tpch-dbgen)
Run dbgen with scale factor 2 (approx. 2 gb to match tpcc with scale factor 2)
Copy all resulting *.tbl files to /srv/data-- everyone should have full access to this folder and its contents
Run setup script as above