We are not seeing these restarts at LLO. I suspect it is because the H1 DAQ is writing 20% more data to disk
From the DAQ screen captures, I get the following stats
So the H1 frame-writers are ingesting 8% more data and writing out ~20% more data to disk. Since L1 is right at the limit, I am not surprised that H1 is having issues.
Possible solutions are (maybe) faster DAQ computers and (definitely) less data to frames. I note the L1 science frames (64s long) are just at the 10Mb/s limit while the H1 science frames are 23% over that limit.
L1 may consider not adding the full suite of fast PCAL channels (when that install occurs) to ensure DAQ stability.