Ability to Drop Indexes via API
Allow users to not only add indexes via the API, but also allow them to drop indexes the same way. Also incorporate index creation/deletion into Terraform.
API index family should come with create, drop, modify, hide !
So, your support told me to give feedback here.
We are using Cloud Manager to manage 73 MongoDB Clusters. (Yes 73, we do IoT).
Currently, we are able to add indexes using the API. We want to drop indexes in an automated way.
We create it through API in cloud manager and Atlas (we store all data from all the locations) at the same time. So we want to drop indexes in Cloud Manager and Atlas through API.
What we are really looking for here is a single, consistent process for index management in Atlas. Rather than doing part of the process one way and the rest of the process another way.
Our Change Management procedures require that any new way of implementing a change go through separate scrutiny. (i.e. testing, documentation, approvals, etc.). Since creating an index via the API would be different than creating an index via scripting the change to be run directly connected to the cluster, each process would have to go through separate Change Management approval. The same goes for index deletion. If all the index management can be done via the API, it all only needs to go through approvals once. Plus, we don't have to code something different for managing indexes. It can all be handled in one piece of functionality.
Note that the managed add index capability is specifically in order to do a rolling index build which essentially minimizes the impact of the build on operational performance (similar to a background index build but even more offline by pulling nodes temporarily out of client visibility entirely while building).
Dropping indexes does not have a comparable operational impact, so you can do those with direct MongoDB wire protocol access to the cluster itself (you can do the same with building indexes where a traditional foreground or background index build suffices).
Can you shed any light on what managing this through the Atlas control plane API would do for you that managing it through the cluster directly doesn't?