I noticed that each night after the logrotate cronjob has run, the (new) myslow logfiles are not filled anymore (i.e stay at 0 length). Also sometimes the logswatch process seems to stop functioning after the nightly logrotate (which causes a loss of mta/spam/virus statistics).
A manual stop and start of the logger is needed to fix this.
Looking at /etc/logrotate.d/zimbra, it seems that a kill -HUP of logswatch is done after rotating zimbra.log, but that appearantly is not always enough.
Is this a know bug in 3.1; is there a fix / workaround ?
(I use fedora core 4)