Description of problem: Syslog-ng is not tolerant of configuration errors -- rather than start those elements that are valid, it will not start at all. Version-Release number of selected component (if applicable): syslog-ng-1.6.12-1.fc6 How reproducible: Always Steps to Reproduce: 1. Configure syslog-ng with a partially invalid configuration (see below for some examples). 2. /etc/init.d/syslog-ng start 3. Actual results: Syslog-ng will fail to start, generating an explanation of the problem. Expected results: Syslog-ng should start those destinations / sources / filters that are valid. It should still generate an explanation of what wasn't started, but valid elements should live. Additional info: Examples of invalid configurations include: - destination bogus { udp( "dest_addr" localip(non-existent-addr) port(8913) ); }; In this case, as syslog-ng can't bind to non-existent-addr, only this destination should be disabled (rather than the entire logging system). - log { source(s_sys); destination(non-existent); } Again, the non-existent destination should disable this one log element, not the entire logging system.
syslog-ng doesn't support silently ignoring config options. Their desired behavior for upstream is to error out when configured options won't work.