> With our current policyd 1.82 instances, we run cleanup every hour and > with about a million messages a day for each, that works out fine. Our > retention time is only around 4 - 8 hours for various values, so we are > deliberately keeping the database sizes small (just a few tens of > megabytes).
Hrmmm .... v2 is mostly automated in terms of what it cleans out vs. the various policies you've setup. Stuff like quotas & accounting require tracking of mail, which is heavy on DB entries if you're doing 1 mil+ per day, by heavy I mean 1 record per message (as it tracks all the recipients per message). I am aware of a few installations 5 mil+ that do cleanup once a day and some that do it every hour. Maybe they can reply on this with their findings? v2 does not lock the entire DB, so cleanup taking a few mins or even an hour or so doesn't really impact anything. Try once a day first, it may take a while, then compare it with once an hour and compare. If you have any issues I am more than willing to look into them for you and work to ensure that cleanup is as fast as possible. I was thinking about an automated daemon to cleanup the DB depending on various stats, so no need for cronjobs. -N _______________________________________________ Users mailing list [email protected] http://lists.policyd.org/mailman/listinfo/users
