What is RethinkDB's maintenance status, and how would you weigh adopting it today?

## What actually happened RethinkDB was built by a venture-funded company that could not turn a widely-admired open-source database into a business. In 2016 the company announced it was shutting down. In 2017 the source code was relicensed under the Apache 2.0 license — it had previously been under the AGPL — and stewardship moved to the Linux Foundation, with the Cloud Native Computing Foundation involved in funding that relicensing. Since then the project has been community-maintained: the repository is alive, issues get answered, releases happen, but at a cadence set by volunteers rather than a funded team. This history is why interviewers raise RethinkDB at all outside of a shop that runs it. It is a compact case study in the difference between *technical quality* and *project sustainability*, and a candidate's answer reveals how they evaluate dependencies. ## The technical merits are not in dispute Changefeeds were, and largely still are, the cleanest expression of database-level push: subscribe to a query, receive `{old_val, new_val}` for every subsequent change, including over ordered top-N windows. ReQL is a pleasant embedded query language. The clustering story includes sharding, replication and automatic failover. None of that stopped working when the company did. So the adoption question is not "is it good?" but "what is the cost of depending on it?". ## The dimensions that actually decide it **Security response.** A database is exposed infrastructure. The question is not whether bugs exist but how fast a fix ships and how confident you are that someone is looking. With a small maintainer pool, that answer is "eventually, if someone volunteers" — which for many organizations is where the conversation ends. **Driver and runtime currency.** You do not consume a database directly; you consume its client library for your language, on your runtime version, in your framework. Drivers rot faster than servers. Check whether the driver you need is maintained, whether it supports current language versions, and what your plan is if it stops building. **Operational ecosystem.** Managed hosting, backup tooling, monitoring integrations, container images, and battle-tested runbooks are all thinner than for mainstream stores. Every gap becomes work your team owns forever. **People.** Hiring for it is hard, onboarding takes longer, and the engineer who championed it becomes a single point of failure. That is an organizational risk, not a technical one, and it is usually the decisive one. **Exit cost.** Because ReQL is an embedded DSL rather than a portable query language, migrating away is a rewrite of the data-access layer, not a connection-string change. That asymmetry — cheap to adopt, expensive to leave — is exactly the shape of dependency to be cautious about. ## How to answer the adoption question A credible answer is conditional rather than dogmatic: - If an existing system already runs RethinkDB and is stable, ripping it out is rarely the highest-value work. Contain the risk instead: isolate data access behind an interface, keep a tested restore path, pin and vendor the driver, and monitor upstream for security advisories. - If it is a new system, the default is no. The burden of proof is on the person proposing it, and it must clear the alternatives on a requirement that genuinely cannot be met otherwise. - If real-time push is the product, evaluate mainstream options first: a maintained database's own change-stream or replication-log facility, or an explicit event log beside the database with application-level fan-out. These need more parts than a changefeed, but every part is maintained and staffed. ## The transferable lesson The reason to know this story is not RethinkDB trivia. It is that dependency selection has a sustainability axis independent of quality, and that the axis matters most for components that are hard to replace — databases above all. The right instinct is to ask, of any infrastructure dependency: who is paid to fix this, how would I find out it broke, and what does leaving cost?