> >However, it is entirely possible and indeed workable, I think, to simply > >send the raw syslog message using sockets and avoid a syslog API > >altogether, leaving the remote log server to make sense of it (which has > >already been mentioned by someone else, and is I think the default > >behaviour of syslog - but I'm wiling to be corrected). But smarter heads > >than mine will decide. > > I think a good case can be made for both. The syslog interface gives you a > quick way to get remote logging up and running because it doesn't require > any special programming. On the other hand, having a socket interface gives > you the ability to write whatever crazy program you want to process the log > information. > > On the other hand, I don't want to make the boot parameters too > complicated. There's already too many of them and some are pretty goofy. > What about a dual-pronged approach? The first being updating Anaconda to support the syslog API, the second being to provide a 'logger' command in bootstrap that can be called by kickstart itself, like the one used on FBSD, like this: logger -t kickstart -p local7.warn @loghost "Could not find /install/foobar.rpm to install!" or something like that. Klaus
