On Fri, 2012-03-23 at 08:25 -0400, Dr. Michael J. Chudobiak wrote: > The init system used a series of hard-coded numbers in the init > scripts to judge which services were to be started in which sequence, > which was a horrible mess. Yet, worked well. > You had to make sure the service X's priority of 37 was in between > service Y'x priority of 18 and service Z's priority of 56. Often, I'd configured systems where I had to specifically arrange the order of a series of services so that one started after another, in a precise manner. Else, all but the first one would fail to start, and never recover. > With systemd, you just say things like: > > After=syslog.target network.target auditd.service > and/or > Before=poweroff.service reboot.service halt.service > > Which is MUCH more concise and easy to understand. But much less precise. It's all very well to say "this," "that," and "the other," need to start after "this thing," is easier to set. But if you definitely need "this thing," followed by, "this," followed by "that," followed by "the other," in that precise order. Then the numbered scheme does exactly what you want. For a more specific example some users may require services to start up in the following order, exactly: network (because just about everything else requires it) ntp (because you need precise logging of everything that starts) samba (because it doesn't work if the network starts after it) apache (because it doesn't work if started before the network, and you need samba running before it for some of the files) And the list could go on. As one thing depends on the prior thing, in a a dependent chain. Trying to get half a dozen things to fire up at once doesn't really work, as some things can't start without a previous thing. They don't begin starting up and keep on trying to start while waiting for what they depend on. They start up, once, and succeed or fail. Sure, having to rename files by hand, to change numbers, can be a pain. But it's not beyond the potential of having a drag-and-drop interface to sort services into the required start up order. > The computer figures it all out, instead of the user having to juggle > priority levels. Evidence seems to suggest (with some prior emails about some services starting too soon) that the computer doesn't "figure it out." It's, also, a bit of a fallacy that sequentially reading files from the hard drive to start up in the sequence that they're needed will be slower than trying to access many different files at the same time. It's like trying to read fragmented files, with more seeking than reading. And seeking is slower than reading consecutive blocks. Trying to read the log is a major pain, too, with multiple concurrent services all splattered in together. -- [tim@localhost ~]$ uname -r 2.6.27.25-78.2.56.fc9.i686 Don't send private replies to my address, the mailbox is ignored. I read messages from the public lists. -- users mailing list users@xxxxxxxxxxxxxxxxxxxxxxx To unsubscribe or change subscription options: https://admin.fedoraproject.org/mailman/listinfo/users Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines Have a question? Ask away: http://ask.fedoraproject.org
