Erald Troja wrote: > Hello all, > > We are using Hsphere control panel automation offered > from Parallels with precompiled Qmail binaries. > > Our entry onto the spamdyke /etc/init.d/qmaild script which > is currently running on a CentOS 4.6 is as follows. > > at the very top we define SPAMDYKE and it's configuration file > > SPAMDYKE="/usr/local/bin/spamdyke --config-file /etc/spamdyke/spamdyke.conf" > > further down onto the start portion of /etc/init.d/qmaild we issue (all > in one line) > > tcpserver -v $RRDNSKEY -R -c $TCP_SERVERS $IPLIMIT $RELAYCHKARG -u > $USER_VPOPMAIL -g $GROUP_VCHKPW 0 smtp $SPAMDYKE $RBL qmail-smtpd vchkpw > true cmd5checkpw true 2>&1 | splogger smtpd & > > Our Spamdyke configuration file is as follows. /etc/spamdyke/spamdyke.conf > > log-level=info > graylist-level=always-create-dir > graylist-dir=/var/tmp/spamdyke.graylist.d > graylist-exception-ip-file=/etc/spamdyke/whitelist.conf > graylist-min-secs=1200 > graylist-max-secs=4322000 > reject-unresolvable-rdns=true > reject-empty-rdns=true > > > Our maximum tcpsessioncount is set to 1000. This has been working > fine for when our Qmail server was operating without Spamdyke. > > Recently we've hit the limit of tcpsessioncount twice. I've been > monitoring the log files and this happens slowly but surely. > > I'd like to ask, why, and what can we do to prevent this and make it. > Raising tcpsessioncount is an option, yet I believe we will slowly but > surely reach the limit as well. > > Thank you. >
Try adding: idle-timeout-secs=660 to your configuration file. I'm betting that will fix you up. ;) See http://spamdyke.org/documentation/README.html#TIMEOUTS for details. -- -Eric 'shubes' _______________________________________________ spamdyke-users mailing list spamdyke-users@spamdyke.org http://www.spamdyke.org/mailman/listinfo/spamdyke-users