Fridge database setup

Use PostgreSQL v16 with TimescaleDB

Database layout

• Single database cryostats
• One table per function
  • temperatures
  • states
• Setup benefits:
  • Easy to setup downsampling operations on each individual schema — might want different downsamping for state vs temperatures
  • Easy to query full list of cryostat names (e.g. for temperatures)
  • If there ends up being more than state/temperatures later, can just add a new schema
  • Having everything in a private database means if you want to use it you just start with a fresh one. Also it’s easy to copy JUST the cryostat data

Database users

• Admin / superuser responsible for
• Read-only user viewer (for e.g. grafana / backup)
• Read + write user
  • Unable to delete/update lines

Uploading to the database

• Each cryostat has a python process which exposes a JSON file via HTTP:

temperatures
	"40K" = 57.6
	"4K" = 3.7
states
	"1k" = "heat_switch_cooling"

• A “collection” server handles uploading the data to the database
  • It has a list of cryostats like hpd = 132.163.100.100:4000
  • It scans through the cryostat IPs and uploads the latest data

Python functions needed

• Upload new data
• Create new cryostat tables (+ add permissions)
• Download a time range of data (with downsampling)
• Dump table (for backup/testing purposes)

Other notes

• Run on Proxmox
• Backup with borgmatic
  • https://torsion.org/borgmatic/docs/how-to/backup-your-databases/