fsck has gone missing, again[SOLVED]

idella4 · Last edited by idella4 on Mon Aug 11, 2008 1:38 am; edited 1 time in total

This is totally weird. A new of instance of gentoo with a new kernel. Boot up gets to the point of mounting root. It at least recognises the root device. It attempts to do the file system check before mounting read write mode. Then it loses fsck.
Firstly, it reports fsck, no such file or directory.
Secondly, it reports the super block doesn't seem to be ext3 file system, etc etc..

These two reports are a load of rubbish. It really lost its way.

It then offers the standard option of enter root password to enter a shell or control d to continue.
Control d doesn't continue ofcourse, it reboots. If it continued there might not be a problem.
On entering a shell as root, it is found that fsck.ext3 is happily sitting there, in /sbin.
There is also no problem with the state of the file system anyway. From this gentoo, it can be checked with fsck.ext3 with no problem.

If that's not enough, from the shell as root, if you attempt to actually do the file check, it again reports falsely the message about the superblock.

This renders the system completely unusable since it can't get through bootup.
Does this make sense to anyone??
_________________
idella4@aus

idella4 · Posted: Sun Aug 10, 2008 4:17 am Post subject:

Has no-one else had something similar happen?? The gentoo I'm working from had something similar happen but it was fixed with a one line script as a work around missing fsck. This is twice in a row.
_________________
idella4@aus

cyrillic · Posted: Sun Aug 10, 2008 4:44 am Post subject:

I have never seen this problem on one of my own machines, but as a temporary workaround, you could edit /etc/fstab to change your dump and pass numbers to 0 0 (this would bypass the fsck during boot).

idella4 · Posted: Sun Aug 10, 2008 10:22 am Post subject:

cyrillic, thanks for that, could I hold your attention again re this. I've just made a discovery.

I currently have 2 versions of gentoo, well 3 actually. Two new and one old. Let's forget about the old one and deal with the 2 new.
The one I am currently working from is about 5 weeks old and I had all sorts of trouble getting it to complete. It began life on the partition as ext3, partitioned by Suse-11 at the time. It does what they currently do and created an ext3 file system, inode size 256 bytes in size, using mkfs. I got to booting it in grub and drew the dreaded grub error 2.
The only solution offered from the reference I was given which explained it was to abandon the new system, repartition stipulating the inode size 128 bytes, using mkfs in a console stipulating the option. I currently boot this up using lilo because grub in its wisdom has taken a dislike to 256 inode sized ext3 and refused to touch the kernel made by it.

Enter the latest instance. It is new made using this method. The idea is to boot up using grub. This is because I want to try using xen and there is now way to use lilo by citing the modules title that grub uses to boot up the xen kernel. I found a way on a gentoo document but the utility to fix it just didn't come in a usable form.

I've just copied the kernel over from the newest to this version so as to use grub to boot, because this one gets through bootup but the newest draws this weird error at boot. The 128 byte sized kernel boots up using grub, but it repeats this impassable state at bootup. So it's the state of the kernel that is the source of the problem.

Obviously I've resorted to the lilo bootup to have booted into the system I'm now in. Presumably I could get past this point possibly by using lilo on the newest, but I would have to try it to see.

How do you get a new gentoo system to bootup with grub AND get past the file check without resorting to your above solution?????????????????????????????
Does everyone else use ext3? If so, how???
I tried using reiserfs and abandonned it because of the same issue. The bootup scripts appear to be hard coded into expecting the file system to be ext3 and to use fsck.ext3 to perform a check. What is wrong with using reiserfs. The gentoo bootup scripts make it impossible.

Is there an alternate file system to use to avoid these weird obstructions???
_________________
idella4@aus

cyrillic · Posted: Sun Aug 10, 2008 3:24 pm Post subject:

I am not really sure how to help you on this one, because my machines are configured a bit differently.

First of all, I am a fan of reiser3, so I have not had to deal with those 256byte inode ext3 issues.

Second, I have not been very happy with the latest grub version bumps.

https://bugs.gentoo.org/show_bug.cgi?id=231039

https://bugs.gentoo.org/show_bug.cgi?id=216307

So I have been using grub2 from cvs/svn instead (because the grub-1.96 ebuild doesn't work), and I have been very happy with it.

BTW, I have dump/pass set to 0 0 in /etc/fstab , although 0 1 would work fine too.

idella4 · Posted: Sun Aug 10, 2008 3:39 pm Post subject:

cyrillic, I take it then that you use reiserfs on your home gentoo then??

You have been helpful in informing me of this so far. My newest version is all but paralised all from being made with ext3 128 byte size inodes. Modules simply don't compile properly.

It looks like I might scrap the latest and re-make it in reiser.
_________________
idella4@aus

cyrillic · Posted: Mon Aug 11, 2008 12:24 am Post subject:

idella4 · Posted: Mon Aug 11, 2008 1:37 am Post subject:

cyrillic
ok, thanks for that. Important thing is to have got past the unnecessary file check.
_________________
idella4@aus

bigbangnet · Apprentice Joined: 11 Jun 2007 Posts: 174

I dont know if its the same problem than I but you might take a look at my post. It looks like the same problem that your having. Well, every details are there.