It's not a problem for now, even if we use oplev damping, because the oscillation leaking to the oplev error signal (attached, a tiny bump at 0.32Hz for green curve) is small enough.
Anyway, it seems like this has been going on for the past four weeks (second attachment), and even if it doesn't get worse, oplev people should look at it after the integration test.
Oplev SUM peak-peak was a factor of 10 worse before it settled down to the current state (beginning of the second attachment), but I cannot see the raw data so we cannot say what was going on then.