User name R. Michaels
Log entry time 10:03:04 on April18,2007
Entry number 199494
keyword=trying to use 4 GB filesplit (failed)
After talking with Jian Ping (RC) I decided to use the downtime
to increase the filesplit limit to 4 GB. This would allow us to
take longer runs with more events if desired. To my surprise, this
failed. We've been running for years with the CODA parameter
SPLITMB set at 1950 which causes filesplitting to occur at just
below 2 GB. If I increase this to 2200 or 2500 it causes
filesplitting to occur at 32768 bytes, generating hundreds of
files quickly (?!). This spurious behaviour occurs somewhere
between 2001 and 2200. (I didn't isolate the exact borderline).
This will have to wait for another day. I returned to SPLITMB =
1950. Did a few test runs. Runs 2621 - 2627 are junk or don't exist.