I looked into the two locklosses from November 29th, 01:39UTC and 04:25UTC from during the time where the AS72 Gain was increased from 12dB to 21dB (74457), before it was reverted due to potentially having caused the next two locklosses.
It doesn't seem to me that both locklosses had the same cause, since there seems to be some discrepencies between the two locks as well as between the actual locklosses. Based on looking through the lockloss ndscopes, it looks like the first lockloss was probably caused by the AS72 gain, but I don't think the second one was, or at least wasn't the only cause. Here's a quick breakdown of some differences:
ASC Loops (ASC-01:39UTC, ASC-04:25UTC)
01:39 Lockloss - Signals look normal up to lockloss
04:25 Lockloss - Many signals are much larger than during the previous lock, and some are increasing in the last ~10s in the INMON channel.
LSC Loops (LSC-01:39UTC, LSC-04:25UTC)
01:39 Lockloss - LSC loops look much different than usual
04:25 Lockloss - Loops look pretty normal
SUS MASTER_OUTs (SUS-bothtimes, SUS-01:39UTC, SUS-04:25UTC)
01:39 Lockloss - MASTER_OUT channel values are more or less stable once in NOMINAL_LOW_NOISE
04:25 Lockloss - During last ~15minutes of lock, many MASTER_OUT values double and stay that large up to the lockloss
Sheila, Oli
I was looking for any overflows from AS72 before the locklosses, and there were none, seen here (01:39UTC, 04:25UTC; zoomed out for the sake of better viewing, but I found the time of lockloss using AS_A_NSUM_OUT while zoomed in). The rise in Accumulated Overflows doesn't actually happen until after the locklosses, which is normal.
So we can maybe say that the increase in the AS72 gain was not the cause of these two locklosses.
An extra small note about the 04:25UTC lockloss: Regular oscillations can be seen(ndscope) on ASC-AS_A_NSUM_OUT starting -9s until LL and on LSC-PR_GAIN_OUT starting -11s until LL.
