/usr will not mount

[Luke Morgan]

I have a bit of a problem!

I am running 10.20 on a K class.
When it was rebooted this morning, it got to a point where it was hanging and giving an error message about init respawning the console.

I have since rebooted using the key and got into single user mode.
Following mount -a,
I did fsck on all the lvols that it said had a problem. They all fixed ok except for
/dev/vg00/lvol7 which is /usr
In this directory is the ARMServer daemon, without which I can't seem to use the disks in the AUTOraid, so when going up to multi user mode, fsck gets run automatically on other file systems.

In single user mode, I ran
fsck -F vxfs -o full /dev/vg00/lvol7
and it seemed to finish ok.
But when I then try mount /usr I get
vxfs: mesg012: vx_iget - file system invalid inode number 1
vxfs mount: /dev/vg00/lvol7 is not a vxfs file system.

Having read a few posts, someone suggested using dd to see if the disk has i/o errors, but that doesn't work. I am assuming it is in /usr/bin ?

Have I any choice but to recreate the logical volume and restore the filesystem?
Can I do this without the commands in /usr ?

TIA

Luke

[Mark Grant]

Firstly, it sounds like the file system is a bit broken.

All commands that you need to recreate this should be in /sbin as well as /usr (just change your PATH) for this very reason :)

You might want to try mounting it as an HFS file system unless you are absolutly sure it was a vxfs file system to start with ("mount -t hfs"

[Sunil Sharma_1]

Hi Luke,

Can you just see /etc/fstab to determine what is the type of /usr file system? wheather it is vxfs or hfs. and try to run Full check on /usr file system in single user mode and try to mount that manually.

Sunil

[Luke Morgan]

Hi, thanks for the swift replies.
The contents of /etc/fstab confirm what I already thought.
It is a vxfs file system.
Having run
fsck -F vxfs -o full /dev/vg00/lvol7
it again seems to pass through it ok, but when I go to mount the filesystem, I still get the vxfs: mesg012 error indicated in my original post.

Luke

[Mark Grant · Accepted solution]

Luke,

I think you are going to have to re-create this. All the LVM stuff is still available in /sbin as is "tar" so you should be able to restore all this.

[Robert-Jan Goossens]

Luke,

before you recreate the filesystem try this,

fsck -F vxfs -o full /dev/vg00/rlvol7

Regards,
Robert-Jan

[Luke Morgan]

Ok.
At risk of sounding stupid, do I just need to do the following to recreate my logical volume?

lvremove -f /dev/vg00/lvol7
lvcreate -L 620

Do I need to use
lvcreate -L 620 -n lvol7 vg00
or will it automatically use the same logical volume number?

Thanks

Luke

[Robert-Jan Goossens]

Luke,

check my previuos post,

fsck -F vxfs -o full /dev/vg00/rlvol7

notice the r in rlvol7

Robert-Jan.

[Luke Morgan]

Robert-Jan,

I have done the fsck command on both the lvol and rlvol and both complete just fine.
When I then try to mount the filesystem however, I get the same error as indicated above.

Luke

[Mark Grant]

Luke,

You probably don't actually need to re-create the logical volume. You can probably just re-create the filesystem with "newfs -F vxfs /dev/vg00/rlvol7" and then restore from tape.

On the other hand if you want to go a bit ultra safe, then re-create the logical volume using the "-n" as you said to be sure.

[john korterman]

Hi,
just a suggestion/guess:
if you use etc/fstab for mount, pleae check the entry concerning /usr for suspicious characters.

regards,
John K.

[Luke Morgan]

Thank you all for your help and assistance.
I removed and recreated the logical volume, created a newFS, mounted it, and then restored the data.
It seems to be working ok now.

Luke