Hmm, there might be an issue with socket activation, which only happens when running under systemd. Can you do an lsof on the syslog-ng process? Does it open a socket somewhere in /run? If you invoke logger with explicit -u option where you specify the name of this socket, does it work? On Feb 8, 2017 5:11 PM, "Jordan Ladora" <vicepresjoebiden@gmail.com> wrote:
I've got syslog-ng v3.7.3_7 running on FreeBSD 11-REL as the log server with both FreeBSD 11-REL and CentOS7 clients (the CentOS clients are running v3.5.6-3.el7 as it's the latest version installed by yum epel repo).
It's always worked perfectly on the FreeBSD systems, but a few months ago I had the same problem described here when setting up a CentOS client-
https://serverfault.com/questions/760383/syslog-ng- working-as-foreground-process-but-not-as-daemon
...I followed the fix of uncommenting the "ForwardToSyslog=yes" line in /etc/systemd/journald.conf and it started logging perfectly on CentOS clients, too.
However, in setting up another CentOS7 client yesterday, this fix no longer worked. If I start syslog-ng via-
chkconfig syslog-ng on systemctl restart syslog-ng.service
..it starts and runs and is shown in ps, starts automatically after reboot, but doesn't log to the server. If I start it manually from the terminal with-
nohup /usr/sbin/syslog-ng -F -p /var/run/syslogd.pid &
...then it does work and starts logging to the server immediately. In that case there are two identical (except for pid) lines shown by ps, e.g.-
482 /usr/sbin/syslog-ng -F -p /var/run/syslogd.pid 904 /usr/sbin/syslog-ng -F -p /var/run/syslogd.pid
...but only the one I started from the terminal (nohup...) actually logs.
Any help would be much appreciated.
-jl
____________________________________________________________ __________________ Member info: https://lists.balabit.hu/mailman/listinfo/syslog-ng Documentation: http://www.balabit.com/support/documentation/? product=syslog-ng FAQ: http://www.balabit.com/wiki/syslog-ng-faq