On Wed, Jan 18, 2012 at 03:19:54PM -0800, Bryan Talbot wrote: > My guess is that the port must be for syslog. Maybe there was a bug > in older 1.4.x versions that don't shutdown cleanly in some cases? I > haven't run into that myself.
I have vague memories of something like this, but much older than 1.4.8, and I've just checked the logs since 1.4.8, I can find no such thing. What I suspect instead is that there are a few dead connections to disappeared users which are waiting for their timeout to strike. Since the timeout is huge (1 hour), the connections appear stuck forever while they're not really. Also, it's fairly possible that some users keep a connection open all the day, and then the old process will remain here to serve them. Running "netstat -antp | grep haproxy" would be useful to find what connections are still alive. But for me the UDP ports are just an artefact. They're here just because the old process is still here and needs to log. Regards, Willy
