pgreplay VS simonwillisonblog-backup

pgreplay parses not the WAL Write Ahead Log but the log file: https://github.com/laurenz/pgreplay

From "A PostgreSQL Docker container that automatically upgrades your database" (2023) https://news.ycombinator.com/item?id=36748041 :

pgkit wraps Postgres PITR backup and recovery:

This blog was just to verify that it works with YugabyteDB. Check pgreplay documentation for more, all works the same in YugabyteDB. If you want to capture a workload from connections on multiple database nodes, each one will have their logfile. You can merge them. The Session ID (the 6th field in the csvlog built from start time and backend pid will probably not collide with another one, but you can make it unique by concatenating a node number if you want). The replay connects to one node, but though a HA proxy the connections can be distributed to multiple ones. All depends on what you want to capture and wh you want to replay. Capturing from PostgreSQL and replaying to YugabyteDB is also a good way to check that all works the same without performance regressions.

> I’ve been running that for a couple of years in this repo: https://github.com/simonw/simonwillisonblog-backup - which provides a backup of my blog’s PostgreSQL Django database (first converted to SQLite and then dumped out using sqlite-

I'm curious, what is the reason you chose not to use pgdump, but instead opted to convert to to sqlite and then dump the DB using sqlite-diffable?

On a project I'm working on, I'd like to dump our Postgres schema into individual files for each object (i.e., one file for each table, function, stored proc, etc.), but haven't spent enough time to see if pgdump could actually do that. We're just outputting files by object type for now (one tables, function, and stored procs files).

My personal blog runs on Django + PostgreSQL, and I got fed up of not having a version history of changes I made to my content there.

I solved that by setting up a GitHub repo that mirrors the content from my database to flat files a few times a day and commits any changes.

It's worked out really well so far. It wasn't much trouble to setup and it's now been running for nearly three years, capturing 1400+ changes.

I'd absolutely consider using the same technique for a commercial project in the future:

Latest commits are here: https://github.com/simonw/simonwillisonblog-backup/commits/m...

Workflow is https://github.com/simonw/simonwillisonblog-backup/blob/main...

If you really, really need to be able to see a SQL schema representing the current state, a cheap trick is to run an automation on every deploy that snapshots the schema and writes it to a GitHub repository.

I do a version of that for my own (Django-powered) blog here: https://github.com/simonw/simonwillisonblog-backup/blob/main...

My blog is Django and PostgreSQL on Heroku, but last year I decided I wanted a reliable long-term public backup... so I set up a scheduled GitHub Actions workflow to back it up to a git repository.

Bonus feature: since it runs nightly it gives me diffs if changes I make to my content, including edits to old posts.

The backups are in this repo: https://github.com/simonw/simonwillisonblog-backup