Cosmos DB Stuck in Updating for Over a Month

Hi **Hunter (FIntivio),
**Thankyou for reaching microsoft Q&A!,

Since the resource is stuck in a control plane “Updating” state, there’s no way to unlock scale operations directly right now. However, you can stabilize the workload with a few temporary approaches until the backend state is cleared.

You can reduce pressure at the application layer by throttling or batching incoming requests. If your workload opens many concurrent connections, lowering parallelism and adding retry with exponential backoff will help prevent the M10 tier from getting overwhelmed.

Another option is to scale out at the application tier instead of the database tier. Increasing the number of application instances and introducing caching for frequently accessed data can significantly reduce the load on the cluster. Even a simple cache layer can buy you time and improve performance quickly.

You can also apply traffic shaping or rate limiting at the API or application level to smooth sudden spikes. This helps maintain stability when the database cannot be scaled immediately.

If the situation is urgent and traffic continues to grow, the most effective workaround is to provision a new Mongo vCore cluster at a higher tier and migrate traffic gradually. You can sync data using tools like mongodump/mongorestore or other migration methods, then switch the connection string once the new cluster is ready. This effectively bypasses the locked resource.

If your workload is read-heavy, offloading reads through caching or a secondary setup can further reduce pressure on the primary cluster.

There isn’t a supported way to cancel or override the stuck operation from your side, so these approaches are meant to maintain performance and availability until the platform state is corrected. Let me know your workload pattern and I can suggest the quickest path with minimal changes.