Something seems to be wrong with my file systems, and I want to fsck everything. But I cannot.
The setup consists of 2 hds, carrying 3 raid1 (ext3) file systems (boot, /, swap). OS is up-to-date CentOS 5.
So I boot from CentOS 5.3 dvd in rescue mode, do not mount the file systems, and try to run fsck -y /dev/md0 fsck -y /dev/md1 fsck -y /dev/md2
For each try I get an error message: "Superblock could not be found..." "The device does not seem to contain a valid ext2 filesystem..."
Well, of course not, the filesystem is ext3, not ext2.
I also tried fsck.ext3, but the error messages stay the same.
So, how could I fsck these arrays?
- Jussi
On Wed, Jul 21, 2010 at 9:42 PM, Jussi Hirvi listmember@greenspot.fi wrote:
Something seems to be wrong with my file systems, and I want to fsck everything. But I cannot.
The setup consists of 2 hds, carrying 3 raid1 (ext3) file systems (boot, /, swap). OS is up-to-date CentOS 5.
So I boot from CentOS 5.3 dvd in rescue mode, do not mount the file systems, and try to run fsck -y /dev/md0 fsck -y /dev/md1 fsck -y /dev/md2
For each try I get an error message: "Superblock could not be found..." "The device does not seem to contain a valid ext2 filesystem..."
ext2 is the base of the journaling ext3 FS.
I would suggest you use TestDisk http://www.cgsecurity.org/wiki/TestDisk and see if you can recover your filesystem. The tool is powerful. I have been able to recover disk partitions and the filesystems within when all other tools reported "no disk partition" on the HDD.
Good luck. -- Arun Khan
On Jul 21, 2010, at 12:12 PM, Jussi Hirvi listmember@greenspot.fi wrote:
Something seems to be wrong with my file systems, and I want to fsck everything. But I cannot.
The setup consists of 2 hds, carrying 3 raid1 (ext3) file systems (boot, /, swap). OS is up-to-date CentOS 5.
So I boot from CentOS 5.3 dvd in rescue mode, do not mount the file systems, and try to run fsck -y /dev/md0 fsck -y /dev/md1 fsck -y /dev/md2
For each try I get an error message: "Superblock could not be found..." "The device does not seem to contain a valid ext2 filesystem..."
Well, of course not, the filesystem is ext3, not ext2.
I also tried fsck.ext3, but the error messages stay the same.
So, how could I fsck these arrays?
Are you using lvm on top of the mdraid?
If so you need to fsck the lvs not the mds.
-Ross
Are you using lvm on top of the mdraid?
If so you need to fsck the lvs not the mds.
No lvm's.
I can boot the raid system, and cat /proc/mdstat shows clean arrays.
The automatic fsck reported problems a while ago, and corrected them. When I now try to run certain rsync script, I get kernel panic. I suspect bad blocks, and/or filesystem errors.
I still don't know why fsck from the rescue dvd does not work.
- Jussi
On 07/22/2010 10:38 AM, Jussi Hirvi wrote: ...
I still don't know why fsck from the rescue dvd does not work.
You could try to let the rescue dvd mount the partitions.
Then you can umount them and fsck.
I don't think that the rescue disk assembles the RAID partitions unless they are going to be mounted.
Mogens
Hei,
Ok, as many of you pointed out, the raid arrays did not start automatically on "linux rescue" boot (when not mounting the arrays).
I followed Mogens' advice: I let linux rescue mount the arrays, then checked /etc/mtab and unmounted everything. After that fsck run ok.
(Turned out the filesystems were ok, so my problem must lie elsewhere.)
- Jussi
On 22.7.2010 11.54, Mogens Kjaer wrote:
You could try to let the rescue dvd mount the partitions.
Then you can umount them and fsck.
I don't think that the rescue disk assembles the RAID partitions unless they are going to be mounted.
Mogens
On 07/21/2010 09:12 AM, Jussi Hirvi wrote:
So I boot from CentOS 5.3 dvd in rescue mode, do not mount the file systems, and try to run fsck -y /dev/md0 fsck -y /dev/md1 fsck -y /dev/md2
For each try I get an error message: "Superblock could not be found..." "The device does not seem to contain a valid ext2 filesystem..."
Perhaps they weren't started automatically? Try running "mdadm -IRs" and then look at the contents of /proc/mdstat and the output of 'dmesg' for more information.
Jussi Hirvi schrieb:
Something seems to be wrong with my file systems, and I want to fsck everything. But I cannot.
The setup consists of 2 hds, carrying 3 raid1 (ext3) file systems (boot, /, swap). OS is up-to-date CentOS 5.
So I boot from CentOS 5.3 dvd in rescue mode, do not mount the file systems, and try to run fsck -y /dev/md0 fsck -y /dev/md1 fsck -y /dev/md2
you should check, using "cat /proc/mdstat ", whether the raid devices were found and started. If not, you could do it manually, e.g. mdadm /dev/md0 --assemble /dev/sdaX /dev/sdbY where the disk (e.g. a and b) and partition (e.g. X and Y) names need to be filled in. There's also mdadm --examine /dev/sdaX , and of course fdisk -l /dev/sda
Once the devices appear correctly in /proc/mdstat, you should be able to fsck them.
HTH,
Kay
For each try I get an error message: "Superblock could not be found..." "The device does not seem to contain a valid ext2 filesystem..."
Well, of course not, the filesystem is ext3, not ext2.
I also tried fsck.ext3, but the error messages stay the same.
So, how could I fsck these arrays?
- Jussi
-
Arun Khan -
Gordon Messmer -
Jussi Hirvi -
Jussi Hirvi -
Kay Diederichs -
Mogens Kjaer -
Ross Walker