This one time, at band camp, Peter Eisentraut said: > Package: clamav-daemon > Severity: normal > > The init script implements the reload action by sending the HUP signal > to the daemon and the force-reload action by restarting the daemon. > That is incorrect. > > The correct behavior, according to policy and LSB, is: > > * If the daemon supports reloading the configuration without restarting, > then force-reload should be the same as reload. > > * If the daemon does not support reloading the configuration without > restarting, then force-reload should be the same as restart, and > reload should do nothing. > > I assume the former is the situation here, so force-reload should do the > same as plain reload. > > (The same issue exists in the init script for clamav-freshclam.)
My undertanding of the proper action for reload is to close open file descriptors, reread configuration files, and reopen file descriptors. clamd only closes and reopens it's log file on HUP, so it seemed to me that it failed the test for 'supports reload'. If you really wanted to reread the configuration, you would need to stop and then start clamd, hence the choice of force-reload being the same as restart. When I was setting this up, it seemed to me to be a useful shorthand to let 'reload' reopen the log file (so that things like logrotate would have a useful target). If that is wrong, I can change it. Thanks for reporting, -- ----------------------------------------------------------------- | ,''`. Stephen Gran | | : :' : [EMAIL PROTECTED] | | `. `' Debian user, admin, and developer | | `- http://www.debian.org | -----------------------------------------------------------------
pgpNUkPROCIDg.pgp
Description: PGP signature
