[twunde]

For new databases that I'm interested in, I generally check if there's a Jepsen analysis for it (https://jepsen.io/analyses) or whether it's being open-sourced by a company that's been using it in production for a while like Foundation DB (https://news.ycombinator.com/item?id=16877395). In either of these two cases I would say it's well-tested. Otherwise, it's often hit or miss. Thankfully the number of new db products I've seen is fewer than the halcyon days of the MongoDB wave (2011-2014) and tend to be a bit more focused on specific problems that the db solves.

[pritambaral]

Jepsen only tests distributed operation. For non-distributed databases (like PostgreSQL, which is — for the purposes of this discussion — what TimescaleDB is), Jepsen does not apply.