Cloud EventStore "Reset" cluster

I have set up a cloud cluster which allows all of our test environments to use.

Periodically, we want to reset the stores backing the longer lived environments, like dev or stage. Is there a way to completely purge a cloud cluster? I know this can be done with deleting files on hosted instances, but I can’t see how to do this on the Cloud ES.


There’s no way to delete the dataset from an existing cloud cluster. What is the use case for that compared to killing the cluster you already have and creating an identical one? It can (and should) be fully automated, and you can store the connection string in your cloud secret manager, so it becomes available for those who use that environment.

Thanks @alexey.zimarev

It’s the time it takes to do the reset, and the person that would do it. If we have a QA environment, I’d like our QA’s to own the ability to press reset and they have a fresh environment. I can easily do this with our database using a simple QA-only backend that they can hit (a restful reset call which performs a db table truncate or similar)

To do this same here, it seems I would have to write some infrastructure automation to provision a new cluster, kill the old one, and change the connection strings. While definitely possible it’s not a quick reset and requires a fair bit of automation on our side to do the infrastructure orchestration - or to do it manually.

We’re anti manual work :slight_smile:

We can’t really bring this feature to the core database as it would require adding things like audit logs. We don’t want to be responsible for someone deleting the database by accident without leaving any trace.

In the cloud it is theoretically possible by shutting down the whole cluster, wiping out the data volume, and starting the cluster again. Unfortunately, we don’t have enough automation to orchestrate complex processes like this in ES Cloud. I believe when we get to the point when we can do more obvious things like automated vertical upgrades, the “cleanup” feature would be possible to implement as we’d have all the necessary automation in place.