Summary
The server code currently stores a cached count of the total number of documents in each collection, for fast access. Certain situations involving unclean shutdowns (when the server crashes) can leave these fast-count values inaccurate. This project will develop an idea to improve this logic in order to avoid inaccuracies in the fast-counts for collections.
Motivation
Accurate fast counts are important, since they are not only consumed by user queries but also internal processes (e.g. sharding). We currently document that fast counts can be inaccurate, but for the users that happen to miss this unexpected detail, it's a form of data corruption.
Cast of Characters
- Product Owner: Michael Gargiulo
- Project Lead:
- Program Manager: Connie Chen
- Drivers Contact:
Documentation
Scope Document
Technical Design Document