CapacityNotActive for just one workspace

I’m seeing the same error and I believe I found a scenario that can trigger it.

In my case the sequence was:

1. Workspace was assigned to Capacity A

2. I created a new Capacity B

3. I deleted Capacity A

4. Only after that I tried to reassign the workspace to Capacity B

The reassignment appears to work. In the workspace settings I can select Capacity B and the workspace shows as correctly assigned to it.

However, some objects inside the workspace (especially semantic models) start returning CapacityNotActive errors. So externally everything looks correct (the capacity is active and the workspace is assigned to it) but internally it seems some artifacts may still reference the old capacity.

Interestingly, the issue sometimes resolves itself after a few days, which makes me suspect there may be some background repair or cleanup process running on the service side.

My hypothesis is that the workspace reassignment expects both capacities to exist during the migration process. If the original capacity is deleted first, the workspace might end up in an inconsistent state.

For anyone doing capacity migrations, I would strongly recommend keeping both capacities active, performing the reassignment while both exist, and only deleting or stopping the old capacity afterwards.