Hi guys, Sorry about that syslog-ng thing slipping in last night. I honestly thought it had been mentioned and sort of discussed already. I realize now that I planned on doing so, but then I worked on the testing and unstable branches and the discussion slipped my mind, but I was thinking all along I had done so. Apologies.
The reasons were outlined by Steve already. The async issue is one of the more detrimental ones. You can easily check this yourself. An easy check is a situation where you have an MTA running locally and use fetchmail to download emails from somewhere that fetchmails hands to your MTA for local delivery. 1) Edit sysklogd's version of /etc/syslog.conf and remove the dashes before the filenames. 2) Download your emails 3) Put the dashes back 4) Download some emails again Without the asynchronous writes, the MTA ends up having to wait for syslogd to write the log entries to disk before the system can continue with the next email. Say you have a hundred emails waiting to download (not such an exaggeration for people who turn their computers off at night, then after a day or so download all their emails including a day's worth of LFS), you will notice that it takes longer for those things to come through. This of course depends heavily on your system configuration, the load it is under already, etc. I've seen cases where email throughput went from 10 emails/sec with sysklogd down to 2 emails/ec with syslog-ng. Personally, that's quite a difference to become concerned with. -- Gerard Beekmans /* If Linux doesn't have the solution, you have the wrong problem */ -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
