For unknown reasons, it happens that Burrito locks a layer and never unlocks it (probably because a pod was killed due to a spot instance reclaim). That causes all the next runs to fail, as the layer is locked.
To prevent this behavior, we could implement a way to unlock the layer before starting a plan / apply operation only if the following conditions are met:
The layer was locked by a Burrito run
The pod that locked the layer is not running anymore
For unknown reasons, it happens that Burrito locks a layer and never unlocks it (probably because a pod was killed due to a spot instance reclaim). That causes all the next runs to fail, as the layer is locked. To prevent this behavior, we could implement a way to unlock the layer before starting a plan / apply operation only if the following conditions are met: