Missing init.d/rcS ??

[amh]

Hi

Finally got my new Gentoo (Stage 3 install) to boot with Grub. However I get a "Busybox" system - whatever that is?

The last console message from the boot process is that /etc/init.d/rcS cannot be found.

I rebooted from livecd to check and that file does not exist on my system.

I'm sure the boot process should not end at the busybox? It appears that none of the init.d scripts have executed at bootup.

Suggestions please....

Thanks Alan

[FreeFly42]

This means init was never installed on your system. For a stage 3 installation this should happen when you expand the tarball. Are you sure you expanded a stage 3 tarball, and not a stage 1 or 2 by accident? Were any errors reported when you expanded the tarball? The good news is you can just reboot from the live cd, mount your gentoo root partition (/mnt/gentoo), and redo the tar expansion. If you've already compiled your kernel you won't need to do this again. Just mount /mnt/gentoo, re-expand the tar, and reboot.

[amh]

Thanks Kent...i'll try tonight. I didn't notice any problems with tarball expansion and yes it was definitely stage3.

Regards Alan

[amh]

Well i've looked into the tar file and /etc/init.d/rcS does not exist in the tar. THe file i'm using is stage3-pentium3-20030910.tar.bz2

I've also checked the stage 1 and stage 2 tarballs and rcS doesn't exist in those either.

Anyone have anymore suggestions please.

Thanks

[darksarin]

I had a very similar problem, and also the init.rd file was NOT in the tarball--I know that I searched it pretty carefully.

[FreeFly42]

Sorry, I was thinking of /sbin/rc. Do you have anything in your /etc/init.d/ directory? Have you emerged anything besides the kernel sources?

[darksarin]

I have to say its been several months since I've tried, and I don't feel like messing with my windows install. I may install mdk 9.2 first to get a working XF86Config file (yes, I know that's cheating), and grub.conf file (so I don't have to guess at the appropriate configs for the most part.

As far as answering your question, I think that the init.d folder was empty or non-existent (can't remember which, sorry).

[FreeFly42]

I hope your /etc/init.d/ folder was empty. The init process looks for scripts in /etc/init.d to execute at startup, and it wasn't complaining about not finding any particular script, but just that there was nothing for it to do, so you were left with a bash prompt.

Most of the scripts in /etc/init.d are put in place by the baselayout package, which I would have thought was installed by the stage 3 tarball, the rest by specific packages as you emerge them. For example, metalog or syslog should have put an init script in /etc/init.d/ when you emerged them.

(Edit: for a stage 1 or stage 2 installation the scripts are installed when you do 'emerge system', so they won't exist in either of these tarballs, which is why I wanted to make sure you didn't extract one of these by mistake.)

You should try 'emerge -k baselayout' and see if it helps.

[FreeFly42]

Just on a whim, I downloaded a stage 3 tarball, and sure enough there in /etc/init.d are all the scripts I would expect from an initial layout.

Code: Select all

khulick@oraxpn2431 download $ tar -tjf stage3-pentium3-20030910.tar.bz2 | grep "/etc/init.d/"
./etc/init.d/
./etc/init.d/nscd
./etc/init.d/sshd
./etc/init.d/keymaps
./etc/init.d/runscript.sh
./etc/init.d/depscan.sh
./etc/init.d/functions.sh
./etc/init.d/clock
./etc/init.d/local
./etc/init.d/reboot.sh
./etc/init.d/domainname
./etc/init.d/localmount
./etc/init.d/consolefont
./etc/init.d/modules
./etc/init.d/shutdown.sh
./etc/init.d/hdparm
./etc/init.d/crypto-loop
./etc/init.d/hostname
./etc/init.d/halt.sh
./etc/init.d/net.lo
./etc/init.d/net.eth0
./etc/init.d/serial
./etc/init.d/rsyncd
./etc/init.d/switch
./etc/init.d/netmount
./etc/init.d/urandom
./etc/init.d/checkroot
./etc/init.d/numlock
./etc/init.d/rmnologin
./etc/init.d/bootmisc
./etc/init.d/checkfs

I recommend you redownload the tarball and expand it again, per the previous instruction, and watch for errors on the expansion.

[darksarin]

I will remember that for the future, but I am not currently attempting to install.

IF you know of a good tutorial on installing grub (with 2 hard drives, and winxp on the primary), let me know. I would like to retry. As a note, if I do this and use this layout:

Code: Select all

hda1=windows xp
hdb1=/boot
hdb2=/root
hdb3=/swap   and
hdb4=/home

then what should my grub.conf look like?

Thnaks.

(Admins, feel free to move this to another thread/forum, just let me know--okay?)

[Corw|n of Amber]

You should not use directly hda4; that should be a primary partition to hold the extended partitions (extended = hdaX where X >= 5).

If you install grub on the MBR of disk 1, your grub.conf will look like this :

#winXP
root(hd0,0)
chainloader +1
#gentoo
root(hd1,0)
kernel=(hd1,0)/boot/kernel-VERSION root=/dev/hdb2

[darksarin]

Yeah, I kinda knew that about the partioning scheme, but I was more interested in the configuration file.

Let me ask one more annoying question--what fs do you use? I have had some good experiences with reiser, but also some bad.

[amh]

Kent

I notice that your listing from the tarball also does NOT have the missing rcS file in /etc/init.d

I have all the other scripts correctly installed - its just the rcS script. If this is a generic script could someone just email/attach it for me.

Thanks Alan

[FreeFly42]

Note: I was interrupted while editing this, so there have been a whole bunch of posts since I started....

darksarin:

I am assuming you would want GRUB as your primary bootloader. If you have something in the BIOS to specifiy which drive to boot from, I recommend not using it as it will cause some goofiness later, though people certainly have done it successfully. My recommended approach for installing GRUB in the configuration you described would be
First, backup the MBR for the Windows drive:

Code: Select all

dd if=/dev/hda of=mbr bs=512 count=1

This will give you a file called "mbr" which you should copy to a floppy or somewhere safe (cdrom?).
Now you can install grub with the commands:

Code: Select all

# grub
grub> root (hd1,0)
grub> install (hd0)

Your Grub.conf would look something like:

Code: Select all

timeout 5
default 0
splashimage=(hd1,0)/boot/grub/splash.xpm.gz

title Gentoo
    root (hd1,0)
    kernel (hd1,0)/boot/kernel-2.4.22-gentoo-r5 root=/dev/hdb2

title Stinking Lousy Unreliable Crash-o-matic Windows
chainloader (hd0,0)+1

You can use /dev/hda4 if you want. You can use any of the primary partitions as an extended partition, though I think you can only have one). If you use all the primary partitions as real data partitions, it simply means you will not be able to have more than 4. If at some point in the future you decide to do something else with your partitions, you can repartition /dev/hda4 as extended and add /dev/hda5 with the same partition boundaries as /dev/hda4 and you won't affect the data at all. Of course there would be no point to this, because you will have taken up all your space with the existing partitions. So you'll have to resize those partitions, which is waaay out of scope for this discussion. Anyway, you can use partitions 1-4 without any problems. And it's simpler to set up, so you might as well.

[FreeFly42]

I notice that your listing from the tarball also does NOT have the missing rcS file in /etc/init.d

I have all the other scripts correctly installed - its just the rcS script. If this is a generic script could someone just email/attach it for me.

I'm sorry, I didn't make myself clear earlier. Your system isn't looking for a particular script called rcS, it's not finding any scripts and just using that as a generic reference. There is no "rcS" script in my /etc/init.d folder either.

So if you have scripts in your /etc/init.d folder, the question becomes why isn't init able to read them? That is truly bizzarre. You wouldn't happen to be trying to have a separate partition for /etc, would you? Obviously your kernel is able to mount your / partition else it wouldn't be able to start /sbin/init...

[darksarin]

Cool, thanks--that's a big help. I have to ask one more question, since I've never encountered a way of backing up the MBR (which is WAY more useful than you know in my situation (or maybe you do know...)). How, then do you restore it?

I would guess something like:

Code: Select all

dd if=mbr of=/dev/hda bs=512 count=1

BUT I am not certain. Thanks in advance.


darksarin

[FreeFly42]

Close. The command you gave would work, but it would also overwrite your partition table. To restore only the bootloader you should only write 446 bytes. Of course since you are going to install on a separate drive your partition map shouldn't change. But Grub won't change that area and it's probably better to just leave it alone!

Code: Select all

dd if=mbr of=/dev/hda bs=446 count=1

There's even a version of dd for windows now, so you could do this backup before you've even started your Linux installation.

[amh]

Kent

Gotya...i understand about rcS (finally). But yes the mystery deepens - all the scipts are their in /etc/init.d. No i'm not doing anything special with partitions - i'm following the recommendations in the install guide. 3 partiitons - boot, swap and root.

This has got me stumped!! I can call the scripts from the command line because i've been starting sshd so i can do the install from another pc via ssh. Here are the exact boot msgs from the console:

init started: Busybox v1.00-pre3 multi-call binary
starting pid 10, console /dev/vc1: '/etc/init.d/rcS'
Bummer, could not run '/etc/init.d/rcS': No such file or directory
Press enter to activate this console

Thanks so far for your assistance....

[darksarin]

Well, you never know how these things work. As I understand it, the easiest way to do this would be to boot from the live CD (both for back up and restore procedures), since if I get grub wrong, then I wouldn't be able to boot windows at all.

I note that you refer to winxp as crash o matic. I have to say there are currently only two or three programs I can't replace. ONe is Pampered Partner (my wife is a consultant for Pampered Chef), the other is SPSS/SAS. My wife also uses Picture It! Publishing, which only runs under windows.

thanks though

[FreeFly42]

amh:

I really thought for sure you were going to come back and say, "why yes, I have a separate etc partition, is this bad?" Rats. Do you have any options on your kernel line in your grub.conf (or lilo.conf if you're using Lilo)?

darksarin:

Yes, you would need to restore it from the installation disk, and would be just fine to take the snapshot from it also. Have you tried running your Windows dependancies under Wine? I am forced to live with an Outlook 5 email system at work, and run Office in Linux using Crossover (Word, Excel, Powerpoint, IE--even with Java and flash (yes I really hate IE, but my stupid company switched the time billing system to use Web-based Project 2000 which requires IE--what's the point of making it web-based if you cram a browser down our throat?), Outlook, all work nicely). I have also used it to run numerous "unsupported" Windows applications in Linux. In fact just the other day I had to teach someone how to ssh to a Linux box from Windows so I ran Putty under Wine on my Gentoo box and connected back to it's own ssh server, just as a demo. Kinda funny.

[FreeFly42]

Also at amh:

I've seen something strikeingly similar to this once. A guy was installing Gentoo on a SCSI-based system, and the LiveCD had trouble loading/configuring the aic7xxx module properly for his SCSI controller, so it couldn't load init. However, the kernel would boot (using the BIOS driver) and would run ok until it loaded the module, etc, then it couldn't access the drives any more. So he got dumped into a Busybox just like you.

This leads me to wonder if you have your hard drive controllers configured correctly in your kernel configuration?

[darksarin]

Yes I have tried. I tried pampered partner under crossover (not plain wine though, since it depends on .mdb files to do its dirty work--its probably nothing more than a vb app), and the picture it failed to install also.


Any ideas?

[amh]

Kent

Hard to believe it is a controller problem - this is a stock standard Dell of about 4 years vintage that has run standard Debian for the first 2 of those years - Intel 440BX chipset - standard IDE controllers etc.

Last question before I give up!! How can i capture all the boot meesages to review closely for any clues - they scroll off console too quickly and there are no files in /var/log where i expected this stuff to be captured.

Regards Alan

[FreeFly42]

Use dmesg. You can pipe it into less for easier searching and navigating, or just send it to a file

Code: Select all

dmesg | less
dmesg > foo.txt

Just because it works in another Distro with their standard kernel (which includes everything under the sun) doesn't mean it will work with your particular kernel configuration. How many people have you seen in these forums with stuff that worked under the live CD but not after first boot?

[darksarin]

me. and with other distros, etc. All in all, I find it fairly frustrating. I realize that some of it is due to the customization level available with gentoo, but I would really love to have an automated installation. In fact, I think that that OPTION would be perfect for the gentoo philosophy--choice and customization. Just a thought, but one that will probably be flamed beyond recognition.

To be honest, I've never understood why they don't have this (especially now that GRP is available). I have heard arguments about Gentoo being about learning linux as well as one can, but I think that the Gentoo Philosophy statement indicates that it is more about choice and flexibility, which is something that I think would be provided by an automated installation.

Now, would I use it? Probably once only, then I would do all others by hand.

Just some thoughts....