Currently, pegasus-dagman sets up the monitord log file and then when launching the monitord sets it's stdout to that log file. However monitord itself takes a backup of the log when it starts up, as it sees that the file exists.

Because of this we always have two monitord.log files in a normal run

Usually for successful run monitord.log.000 is a zero byte file.
rw-rr- 1 vahi gridstaff 718 Jan 29 18:33 ../creamce-sharedfs/work/vahi/pegasus/blackdiamond/run0022/monitord.log
rw-rr- 1 vahi gridstaff 0 Jan 29 18:19 ../creamce-sharedfs/work/vahi/pegasus/blackdiamond/run0022/monitord.log.000

However, for workflows where pegasus-dagman restarts monitord this gets messed up, and we see partial logs from same run in two log files