Ceph

So - turns out - i installed the new dashboard module and enabled it, I guess it caused a reload or swap of the mgr (since there is a small gap in the prometheus metrics).
Anyway, after enabling the dashboard the number of objects and used pool size dropped back to the previous values.

So, whatever it was, it has magically fixed itself. It might be a prometheus exporter issue, though I'm pretty sure I have seen the wrong amount of used TiB in the autoscaler output as well but haven't screenshotted it.