How to use streaming replication with PostgreSQL

In this post, I want to explain how to set up streaming replication for PostgreSQL. We want to be able to do (read-only) reporting on live data, but reduce load on the main database server. Some background: We offer Metabase to many of our customers for data analysis and reporting. This means we don’t have to hand-code everything into our application. It even allows savvy customers to write their own reports.

Kernel backups for Debian security updates

Two of our most important machines, both running Debian Stable, failed to boot after a routine kernel security update. In both cases, the security update itself was not the cause of the boot failure, but merely triggered latent problems (for the full story, see below under The problem with the snapshot machine). Also in both cases, it would have been a great help if we could have booted the old kernel/initrd.