Since you’re on btrfs, a copy-on-write file system, you should also investigate disabling full page writes [1] for Postgres WAL. I disabled it for our Postgres

Since you’re on btrfs, a copy-on-write file system, you should also investigate disabling full page writes [1] for Postgres WAL. I disabled it for our Postgres cluster on ZFS and got a nice reduction in write volume. [1]: https://www.2ndquadrant.com/en/blog/on-the-impact-of-full-page-writes/ https://www.2ndquadrant.com/en/blog/on-the-impact-of-full-pa...