Package: aide-common Version: 0.16-1 Severity: wishlist Hi,
please consider using dotlockfile -p -r 0 -l $LOCKFILE in the wrapper. The current setting will retry to obtain the lock, which is probably not intended behavior if a manual aide run prevents the cron-job from running. I have lost my aide database in the following situation: - I boot up a VM that was not running during cron.daily time - I upgrade the kernel - I reboot before anacron starts cron.daily - After the reboot, I invoke aide.wrapper --update to refresh the database - during this operation, anacron starts cron.daily, aide's cron job cannot obtain the lock because the manually started aide cron job holds it. the cron.daily wrapper waits. - The manual aide run ends, the cron.daily aide run obtains the lock and begins running. In this process, aide.db.new is truncated to zero size. - I review the log from the manual aide run, find it ok, and copy the (zero length) aide.db.new to aide.db. => boom, database lost. I think it is the lesser evil to not have the cron job wait for the aide lock and have it bomb out immediately. Greetings Marc
