No comment on the potential issue.... just a 'me too'. I'd love to see a 'processed' count. My syslog server is processing about 10M lines/day, by my best estimate, but would like to know more precisely. A byte count would be nice, too; currently about 1.2GB day. I'd like to generate RRD graphs, if I could get that data. I tried writing a couple programs to do line & byte counts, and set it up as a 'program' destination, but it would sometimes drop stuff when it looked away from it's input long enough to write the stats to a log. Probably when the there was a burst well above the ~ 115 messages/sec average. User-specified destinations & macro/template formating for stats output would be fantastic!
Stats: dropped is great, would this be better?: Stats: x processed y buffered/queued z dropped
* I'm seeing syslog-ng process sizes in 20-30MB and am wondering if my queue sizes should be upped.