https://issues.apache.org/SpamAssassin/show_bug.cgi?id=7127
--- Comment #17 from RW <[email protected]> --- (In reply to Reindl Harald from comment #8) > bayes_learn_to_journal 1 You've said before that you don't do autotraining, and this setting does nothing useful except reduce the impact of autotraining on Berkeley databases. It's a rarely used feature because anyone that might benefit from it is unlikely to be using a Berkeley database in the first place. This explains why --no-sync could make a difference. I suspect that you problem will go away if you unset this. -- You are receiving this mail because: You are the assignee for the bug.
