Package: zabbix Version: 1:1.1-5 I'm using zabbix backported to Sarge (made by myself). I found both agent and server sometimes fail to restart during log rotation, because tcp port they listen is still in use (in TIME_WAIT state). It happen about one on ten restarts of agnet or daemon (when restarted mannually).
Output from zabbix_agent.log: Cannot bind to port 10050. Error [Address already in use]. Another zabbix_agentd already running ? netstat shows: netstat -an | grep 10050 tcp 0 0 192.168.101.20:10050 192.168.101.20:32848 TIME_WAIT I found this on Zabbix Forum: http://www.zabbix.com/forum/printthread.php?t=1826 Problem could be fixed by adding this line after stopping the zabbix_agent in init file: while [ `netstat -an | grep 10050 | wc -l` -ne 0 ]; do :; done and this line after stopping zabbix_server: while [ `netstat -an | grep 10051 | wc -l` -ne 0 ]; do :; done assuming, that agent and server listen on default port. I suggest adding some kind of detection if tcp port, agent or server is using a) before agent or server start or b) after agent or server is stopped My system: Debian Sarge Linux monitor.prz.edu.pl 2.6.16-2-vserver-686 #2 SMP Tue May 23 07:55:55 CEST 2006 i686 GNU/Linux -- Krzysztof Raczkowski Rzeszow University of Technology System Administrator of CZ RMSK tel.: (017) 865-13-93; e-mail: [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
