The 2004-02-28 at 18:35 +0200, Hylton Conacher (ZR1HPC) wrote:
At that point, only "/" is mounted, not "/usr" nor "/var" - it may need something in there. It is not the first time they put something in /usr that is needed during the manual fsck part. So basically the program needed might not be accessible from the hdb partitions. GREAT :(
Note: _might_ - ie, not necesarily "will not".
OK, my new understanding so far ie: 1. Don't fsck a mounted partition 2. Boot from a CD/floppy to get cli access
Correct. Better from the CD, less work.
Ok, I don't have SuSE 9.0, only 8.2, but I can guess what that rescue entry is. Not good enough in this case. I dunno, but it sure as hell is on the lilo menu if I boot from CD.
You need to boot the rescue system, from the CD or DVD, not from the HD. OK that's easy enough ie insert CD 1 and reboot...np.
Get to the bash command prompt there.... HOW?? When the machine reboots I am presented with a lilo graphical menu enabling me to either Boot from HDD, Installation, Installation with ACPI disabled, Install - Safe settings, Manual install, Rescue System, Memory Test. There is no option to go into a CLI mode???
Select "Rescue system", while booting from the CD.
Choosing the Rescue System option the system starts asking me about my kb type. Right there and then I exited out of it as it looked too much like install, and asides you said to get into a CLI.
Actually, it is probably the same thing as the install part, with a different config or init or whatever. The rescue system has to boot from CD a system it knows nothing about, using a CD... exactly the same as it does for installing. No wonder it asks for the keyboard. You should get to the CLI after one or two selections (language...). Another way to get there was (time ago) to start the install, then press "ESC" - or even ctrl-alt-F2 to it.
....., and issue the fsck command plus options there. Something like "fsck /dev/hdb1". OK understood, but getting to the CLI from CD 1 seems to be the problem ie is there some special Susian key combination needed to access it or something...like Ctrl-Alt-Del :)
No :-) Anyways, the rescue system must be explained on the book - it is chapter 11, in suse 8.2
Asides from that then how am I going to be able to issue a fsck command like -VAa which will step through the fstab file if the fsck doesn't know where to find it?
Ah, you can't. You need to have somethings printed before hand: the partition table, the fstab file... with that paper, you first check the main partition, the one that holds "/"
Now my second idea, your fstab, which I saw on another post. There you have entries for hdb and hdc - but you mention that hdc was later removed. If there are entries for hdc (hdc1, for example, a windows mount), but no hdc disk on the system, it will not boot! And it will precisely tell you to fsck manually.
surprise, surprise, the system still boots OK, into linux thankfully. hdc contained Win95 and was moved for testing purposes. It will never be reinstalled into the linux machine. I was wondering about commenting out those entries of hdc or even deleting them.
Maybe 9.0 is cleverer. Once I had a non booting linux system because I had removed a disk or a partition without editing the fstab file. Perhaps... it could be that windows partitions are not checked.
So, you have to manually mount that disk from inside the rescue CD, edit the fstab (with vi, I don't think there is anything else there), and comment out those lines. KWrite will do it too thankfully when logged into another X session.
That's not fair :-p There is no X system on the rescue system.
How the hell do I get my system back by completing the fsck!!
Well...
The man pages of e2fsck and resiserfsck are printed at the end of the SuSE admin book just for these ocassions. so I see....but there aren't half the options listed there that are listed on the fsck man page. I'll resort to my own written notes about this.
I wrote my notes about this on a handbook, but they are on another city right now. O:-) UPS delivers worldwide I hear :)
Provided there is somebody there with the kind of knowledge needed to find somebody else's book under a pile of them ;-) Heh, I'm right now at that other city, with another older computer: P130-32Mb... Slowwww! But I found one of my notebooks: For ext2 filesystems, I used "e2fsck -f -v -C0 /dev/hda3" or "e2fsck -f -c -v -C0 /dev/hda3" - the second one does a badblock check. Of course, you need to change "hda3" with your device and partition number.
A little informed but still desperate to fsck (defrag?) the system.
No, not defrag. It is equivalent to the windows "chkdsk program"
I am trying to leave the machine on as much as possible here and so far haven't been hit by a power outage....yet. When it does hit I am going to need to know howto use this fsck thing immediately.
So the questions remain: 1. How to get to the CLI from the rescue option when booting off the CD?
Somewhere above.
2. How can I get fsck to walk through my fstab file if the / partition isn't mounted?
You can't. If needed, you can mount it read-only under "/mnt", as "Ken Schneider" said. Otherwise, I prefer to issue "fdisk -l" to list the partitions and have educated guesses.
3. What fsck -option would be the best to check and repair the fs's and also give me verbose output and a progress bar
-v -C0 But one filesystem at a time, manually. Notice, by the way, that if the fstab contain the wrong info, it will force you to a manual fsck instead of an automatic one. -- Cheers, Carlos Robinson