There's a ticket open - http://jira.qos.ch/browse/LBCORE-147 - that it causing me grief. I'd like to offer to have a go at fixing it. The problem as I see it is that if you keep, for example, 7 days worth of log files, then ... On day 8, the appender deletes the log file for day 1 On day 9, the appender deletes the log file for day 2 (Day 10 has no activity) On day 11, the appender deletes the log file for day 4 So, the log file for day 3 hangs around indefinitely until it's tidied up manually. I can see two possible solutions for this; a) Search back for a configurable amount of time, and delete more than one log file. So with the above example with a 1 day search back window, On day 8, the appender deletes the log files for day 1 and day 0 if they exist. On day 9, the appender deletes the log files for day 2 and day 1 if they exist. (Day 10 has no activity) On day 11, the appender deletes the log files for day 4 and day 3 if they exist. b) Check the contents of the directory that the log files are stored in, and delete all those that are older than the required time, so ... On day 8, the appender checks the directory, and finds that the file for day 1 has a name pattern that is over 7 days old, and deletes it. On day 9, the appender checks the directory, and finds that the file for day 2 has a name pattern that is over 7 days old, and deletes it. (Day 10 has no activity) On day 11, the appender checks the directory, and finds that the file for days 3 and 4 have a name pattern that is over 7 days old, and deletes them. I think my preference is probably for option a), but I'd rather spend the time and effort implementing something that is more likely to be accepted as a patch. What do the maintainers think is the correct way to fix this issue? Thanks, Greg