Debugging Firestore Excessive Reads

I’ve been using Firestore for RateRudder because it is simple, free, hosted for me, and supports subcollections. The subcollections are extremely useful in a multi-tenancy schema because every tenant can have their own set of collections under their namespace. Firestore pricing is based on how many documents you read, write, or delete. The free tier includes 50,000 reads, 20,000 writes, and 20,000 deletes per day. The pricing for Writes and deletes are straightforward, but reads are complicated. You’re charged at least 1 read for every document scanned in order to fulfill a query, with a minimum of 1 per query. Additionally, you’re charged 1 read for every 1000 index entries except for specific single-field index queries. You’re also charged for non-document queries, like listing collection IDs. If Firestore happens to be down the Go client will hang waiting for the connection to succeed, or until the context is cancelled. I set up a Cloud Run health check to ensure that a Firestore…