Durability in the case of database catastrophe

Hey Camunders

Just a question around the durability of data within Camunda. We’re looking to setup a clustered environment, and we’d rather not setup a complicated master-master replicated database cluster in the early phases of this project.

Does Camunda have any guards against database unavailability? If we have a master-slave replicated database setup, and the master dies - does Camunda buffer all process-related writes? History is not as important to us right now. I’d imagine the answer would be “no” but I figured it was worth a try.

Many thanks!

Hi @danny_xneelo,

your assumption is right. The database is the single point of truth for the process engine. You have to make it as robust as possible.

There is no buffer between main memory and database.

But it depends highly on your processes how long the data are only in main memory. And if the database breaks, the commands will throw an exception to the caller.

Hope this helps, Ingo

You’re the man, thanks @Ingo_Richtsmeier
Had to try my luck :slight_smile: