On Thu, 10 Mar 2005 21:09:18 -0600, Jason Stafford <jason@dpcomm.net> wrote:
I'm trying to setup a load balanced system with several web servers and one central log server. Currently I have syslog-ng running and pipe the output form apache to logger. This is all running locally on one box. However, syslog-ng just randomly hangs and apache stop processing requests.
This sounds like an issue with the pipe. If you move the logger off-box and switch to UDP, you won't have the problem with Apache hanging, if the remote logger dies, then you just won't have log data.
I can restart apache and it still does not respond, but i can just restart syslog-ng and then it all starts working again. Is syslog-ng designed to handle this kinda of load, around 3 million entries a day?
I have a central logger doing 5 million entries an hour, so I'd say yes, syslog-ng can handle this load :) What sources does syslog-ng accept from on the central logger? TCP? UDP? both? local socket/pipe? I've never used syslog-ng with a local log socket or pipe.
I have been googling for hours and can only find some comments about log_fifo_size and reap_time in the options section?
Currently I have the following: sync(256); log_fifo_size(4095); time_reap(300); log_msg_size(2048);
here is my current config, if that helps any
options { sync (0); time_reopen (10); log_fifo_size (1000); long_hostnames (off); use_dns (no); use_fqdn (no); create_dirs (no); keep_hostname (yes); }; . . . httpd.conf log line CustomLog "|/usr/bin/logger -p local0.info -t APACHE_ACCESS_LOG" tracking
I do not recall if this is the recommended way to have apache log to syslog? Kevin