2.7 Hz is the crossover frequency between TM, IM, and MN in the current hierarchical control.
So, this oscillation is most likely due to the instability of the hierarchical actuators.
There are two possibilities I can think of at this moment:
1. Poor damping performance rotates the phase around 2.7 Hz, which induces the instability at crossover frequency.
2. Relative actuator efficiency mismatch changes the cross-over frequency slightly, and causes the instability.

To confirm if the damping performance affects the instability or not, it is better to try the lock acquisition with high bandwidth damping control of ETMX MNL, which introduces larger noise at high frequency but more robust.
The easiest way to try the high bandwidth control trial is as follows:
1. Request ETMX guardian to ALIGNED state.
2. Edit New_paramsA.py (comment out the lines related to MN_DAMP_L, NBDAMP_L2, and NBDAMP_L3 in the state between ALIGNED and LOCK_ACQUISITION, and between LOCK_ACQUISITION and CALM_DOWN state).
3. Load the guardian.
4. Request LOCK_ACQUISITION state and try ALS_DARM lock.

If the above change will solve the 2.7 Hz oscillations and ALS_DARM becomes stable, the damping update will solve the problem, so the suspension commissioning is necessary.
If not, more fine actuator efficiency tuning between MN, IM, and TM is necessary.