I don't use that haproxy-config.py script (that creates a configfile from a directory layout; maybe some distro ships haproxy with that kind of configuration).
This is how my logging for haproxy looks like: /var/log/haproxy/*.log { daily missingok rotate 365 compress dateext delaycompress notifempty create 640 root adm sharedscripts postrotate /bin/kill -HUP `cat /var/run/rsyslogd.pid 2> /dev/null` 2> /dev/null || true endscript } You don't touch the haproxy daemon at all, so this will work. Alternatively you can use the '*copytruncate' *option of logrotate. -- Best regards, Hauke Hi Hauke, thanks for your answer. One question about the script. What means config in line 20 HAPROXY_CONFIG=/usr/local/sbin/haproxy-config.py / Configurationfile is setting on line 17 Another question is configuration haproxy for logging and logrotate - in logrotate I reload haproxy for creating new logfile - interaction with corosync? *Von:* Hauke [mailto:mail...@haukebruno.de] *Gesendet:* Montag, 25. März 2013 11:41 *An:* haproxy@formilux.org *Betreff:* Re: haproxy in cluster with pacemaker and corosync Hi Wolfgang, maybe it helps: I use the simple lsb RA for haproxy in pacemaker. A modified version of https://github.com/finnlabs/haproxy/blob/master/haproxy-init[1] works for me. Environment: Active/Passive, Debian Squeeze, pacemaker and stuff out of stable repos, self build haproxy (stable). -- Best regards, Hauke On Monday, March 25, 2013 10:22:08 AM Wolfgang Routschka wrote: Hi everybody, today I would like to ask you about haproxy best practice in pacemaker/corosync setups. We want to use pacemaker/corosync for failover haproxy. I found 2 ways for configuration - over lsb-ocf-script and a ocf-script https://raw.github.com/russki/cluster-agents/master/haproxy[2] What´s your "best practice" to configure haproxy in pacemaker/corosync cluster? regards, Wolfgang -------- [1] https://github.com/finnlabs/haproxy/blob/master/haproxy-init [2] https://raw.github.com/russki/cluster-agents/master/haproxy