We were updating some configurations in one of our clusters and then we missed one config option that was required.
This caused our component that updates the apps to be lost in the process of updating the apps, later we tried to fix this issue reverting part of the update but then the permission of this new configuration was lost as well, resulting in another error with permissions.
In the end we added the config option that was required and the scheduler started to work fine again. As a measure to avoid this issue in the future we are going to make this code more reliable and not failing for all apps when only one app or one cluster has problems on its config.
Sorry for the trouble.