When a cycle has an error, it should be broadcasting a cycle error observer message, which should be picked up by the dashboard and putting on the relevant instance's log. This isn't happening, and so the cycle appears as if it's just resetting itself midway.
A reliable way to see this is at the moment (before #93) is to define a backend with a nonsensical style ID.
When a cycle has an error, it should be broadcasting a cycle error observer message, which should be picked up by the dashboard and putting on the relevant instance's log. This isn't happening, and so the cycle appears as if it's just resetting itself midway.
A reliable way to see this is at the moment (before #93) is to define a backend with a nonsensical style ID.