> > > --__--__-- > > Message: 3 > Subject: Re: post installation problems > From: Brian Marsh <b.marsh@xxxxxxxxxxxx> > To: fedora-list@xxxxxxxxxx > Organization: Anansi Ltd > Date: Sun, 22 Feb 2004 15:56:33 +0000 > Reply-To: fedora-list@xxxxxxxxxx > > On Sun, 2004-02-22 at 14:40, Jeff Vian wrote: > > > > > > > > > The sequence of starting things is controlled by the listings in the > > /etc/rcX.d directories. (X is your runlevel) > > > > If you look there you will likely find SXXpcmcia has a higher number > > than SXXnetwork. > > If you change that by renameing the files so the pcmcia starts first it > > will solve this problem. > > BTW, the XX in the names above indicates the sequence they start. Lower > > numbers start first. > > Delaying the start of the network will easily fix this for you. > > > > Thanks for the help, and yep, that all makes sense, except that I have > directories rc0.d through to rc6.d and directories rc0.d, rc1.d and rc.6 > contain KXXetc and not SXXetc. Erm, guess this is my lack of knowledge > of terminology, but what is my runlevel (and how do I determine it is > necassary) and therefore which directories should I delay the network > startup in? Or should I simply delay it in all? > The runlevels are basically "configurations" of the system, with a few exceptions. These are the runlevels that are available by default (this is from /etc/inittab): 0 - halt the system 1 - Single-user mode (provides root only access for maintenance) 2 - multiuser mode with no NFS 3 - multiuser mode with no X Windows 4 - Not used, although you will find files in /etc/rc.d/rc4.d 5 - multiuser mode with X (this is the normal default runlevel 6 - reboot the system You determine your default runlevel by examining the "initdefault" line in /etc/inittab. id:5:initdefault: Normally this is runlevel 5, but sometimes it is runlevel 3. If you run a server, you frequently use 3 to leave X turned off. The exceptions are, of course, 0 and 6, which aren't states you would "run" in. As you probably guessed, anything starting with a K in rcN.d is going to kill the named process, and anything starting with an S is going to start the associated process. So it's appropriate that runlevel 0 kills all processes, and runlevel 6 kills everything except the reboot process. The files in a directory like /etc/rc.d/rc5.d are actually symbolic links to scripts in the directory /etc/rc.d/init.d. Don't change the files in the init.d directory or you can break the system. If you only change the links in rc5.d or its brethren, you might break a runlevel, but you can still use the others and run the system to make repairs. If you are using a vanilla Fedora installation as a workstation, you probably are using runlevel 5. So make the changes in /etc/rc.d/rc5.d and test it out. You can make the changes to other runlevels, in case you use them, but remember that 0 and 6 have their special uses. You don't want to start anything, for instance, in runlevel 0. To do this a little more safely, you will likely want to change initdefault to 3 FIRST. Then reboot and issue the command init 5 to go to runlevel 5. You have to be root to do this. Once you verify that it all works, change the initdefault back to 5. > Don't fancy 'breaking' anything at this stage of experience... I agree with you. Read this over carefully to be sure you are comfortable with this before you try it. For more safety, make sure you have a boot diskette or use the rescue mode on your CD's to be able to boot for repairs. This probably makes it sound a lot more risky than it will be, but discretion _is_ the better part of valor. Another less elegant way to do this is to add a line like service network restart to the file /etc/rc.d/rc.local which is last to run. This restarts the network after everything else is done. But this will make your boots a little longer each time. Erik
Attachment:
signature.asc
Description: This is a digitally signed message part
