Today, IFO could not be recovered to even RF-locked state automatically by the guardian for a long time (more than 1 hour).
Since this is the first time that IFO could not be recovered automatically during O4c observing run, I investigated the reason.

Figure 1 shows the ETMX OpLev values before and after the final lock of the interferometer.
Though pitch and yaw values were recorded well, LS oplev values were changed a lot.
This is probably due to the DARM control feedback signals, so it is hard to control.

To avoid the same instability, what we can do is
1. Record LS Oplev values of ETMX and move suspensions so that LS OpLev values becomes close to the one before lockloss.
2. Move ETMX angle during the lock acquisition so that arm alignment becomes good, for example by ADS using GRX.
3. Decouple L2P and L2Y more to reduce the misalignment when LS OpLev values are changed.

Since there are several advantages and disadvantages in the above 3 countermeasures, we need to consider what kind of treatment should be done.