Javier changed the pedestal suppression threshold resulting in a (slightly) larger average event size -> larger total file size for the 1.4M #events threshold in CODA. Recent runs seem to be generating ~150MB foo.1 file after splitting at 2GB.
We'll change the CODA event limit to 1.3M events (to put the total run size under 2GB again) and see if the analyser behaves in the future.
(Hmm...there must be more to the problem than this. Some of the split
runs replay correctly -- perhaps there is a counter overflow or buffer
overrun when the number of ADC channels with data gets 'large'?)
A copy of this log entry has been emailed to: ole@jlab.org, hrosinko@jlab.org, gomez@jlab.org