This happened again with run 2385.
The real 2385 is in /adaql2/data4, last written at 10:23 (3/5/11).
The fake 2385 (aka 2386) is in adaql2/data1, currently being written.
When the run stops, I will have coda killed, change the filename in data1, and restart coda. Hopefully the RC GUI will start up ready for 2387 and not 2386.