Hi, just to say that I see the same thing. I am trying to copy all the logs to a loghost over udp. And at startup, syslog-ng seems to screw up. On the opposite to the author of that bug report, I do not see anything at all, no connection_broken messages, even after I manually launch syslog-ng after the boot completes.
I have tried to supply both names and numeric ips in the config file and neither work. Adding a static entry in /etc/hosts does not help either. Changing the starting priority is sufficient to fix the problem. It might be related to a bad interaction with nscd (which starts at 20, while syslog-ng starts at 10) since the network should already be up at that stage. Setting a starting priority of 20 or more lets the deamon start. Using the original 10 or 19, the deamon dies. I don't think that nscd should be started earlier. The logging daemon should not need it up to start properly. At worse, it should just delay the name resolution a slight bit. I would also consider this bug upstream and critical, having a logging deamon silently not starting is kind of a show stopper. And this fix of mine is pretty dirty, let's be honest... jacques
signature.asc
Description: Digital signature
