> As an added note I did not see the zombies in 2.5.4 or 2.6.4. Yes, that is normal, the 2.6.4 did not support a warm reload.
(2.6.4) 'amavisd reload' == (2.7.0) 'amavisd restart' For the same reson the 2.7.0 'amavisd restart' would not leave behind any zombies, because the PID of a master process changes from the previous incarnation and zombies are reclaimed by the init process. > I'll try to get some debugging done to help out in my environment. I can reproduce this now. Seems like the Net::Server is blind for the CHLD signal during a short interval after a reload. Manually killing just one of the new child processes is enough to notify Net::Server's SIGCHLD handler to ripe the terminated process along with all the forgotten child processes. The same also happens automatically as soon as one of the new child processes voluntarily terminates on reaching its $max_requests limit, so the situation sanitizes by itself eventually. Looks like some modification to Net::Server will be needed to handle the situation sooner. Mark ------------------------------------------------------------------------------ Learn how Oracle Real Application Clusters (RAC) One Node allows customers to consolidate database storage, standardize their database environment, and, should the need arise, upgrade to a full multi-node Oracle RAC database without downtime or disruption http://p.sf.net/sfu/oracle-sfdevnl _______________________________________________ AMaViS-user mailing list AMaViS-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/amavis-user Please visit http://www.ijs.si/software/amavisd/ regularly For administrativa requests please send email to rainer at openantivirus dot org
