I just updated Fedora 14 with the updates it downloaded and I am now stuck with the error.
Please help... Thank you.
![]()
I just updated Fedora 14 with the updates it downloaded and I am now stuck with the error.
Please help... Thank you.
![]()
Have you typed dmesg as it suggested? Paste the output of that command here so we can look at it with you.
Kurt Driver
Vancouver, Canada
https://www.wikitree.com/wiki/Driver-1479
the output is really long and will not allow me to paste it all here...
Here is a link to a text file that contains the output.
http://dl.dropbox.com/u/8228842/output.txt
OK, here you go... Thank you...
[root@space Desktop]# dmesg |tail
[ 9155.124300] sd 4:0:0:2: [sde] Write Protect is on
[ 9155.124304] sd 4:0:0:2: [sde] Mode Sense: 02 00 80 00
[ 9155.124308] sd 4:0:0:2: [sde] Assuming drive cache: write through
[ 9155.132669] sd 4:0:0:3: [sdf] Attached SCSI removable disk
[ 9155.137793] sd 4:0:0:2: [sde] Assuming drive cache: write through
[ 9155.137801] sde: sde1
[ 9155.152546] sd 4:0:0:0: [sdc] Attached SCSI removable disk
[ 9155.157945] sd 4:0:0:2: [sde] Assuming drive cache: write through
[ 9155.157951] sd 4:0:0:2: [sde] Attached SCSI removable disk
[ 9265.643440] usb 2-2: USB disconnect, address 2
It may amount to nothing, but take a moment to list your running kernel and enabled repos...Maybe no help, but no harm either.Code:uname -r yum repolist
Here is the output for the two commands.
uname -r
2.6.35.11-83.fc14.i686
yum repolist
Loaded plugins: langpacks, presto, refresh-packagekit
Adding en_US to language list
repo id repo name status
adobe-linux-i386 Adobe Systems Incorporated 18
fedora Fedora 14 - i386 17,883
google-chrome google-chrome 3
google-talkplugin google-talkplugin 1
updates Fedora 14 - i386 - Updates 5,793
repolist: 23,698
Ok. Just realized your problem... maybe.
Was the partition you are mounting previously used in a LVM, and reused as a raw partition?
IF so, then this happened to me.
I have worked around it - first by commenting out the filesystem so that the system will boot without it (unless it is the root filesystem), and then mount it manually (I uncommented the fstab entry).
The problem is that when a partition is initialized for LVM use, it marks the partition for such with the string "LVM2_member". Unfortunately, mkfs doesn't overwrite this identifier.
This is also reported by "blkid" utility as a "type=LVM2_member".
My initial fix was crude (but worked).
1. have a backup of the partition data (I used tar), and unmount the filesystem
2. I could not identify any utility to overwrite this so I did a
dd if=/dev/sd<partition> bs=1 skip=512 count=512 | od -cb
This shows the string, and verifies the problem.
3. Get the list of backup superblocks from the filesystem (this works for ext2/3/4)
dumpe2fs /dev/sd<partition> | grep superblock
4. erase the string. My clumsy way was to erase the block (this isn't necessary, as if you identify the exact bytes from step 2 you can erase
just these). I erased the entire block with
dd if=/dev/zero bs=1 seek=512 count=512 of=/dev/sd<partition>
5. Because I erased the entire block, I had to recover using fsck with one of the alternate superblocks:
fsck -y -b <block number from step 3> /dev/sd<partition>
Note: erasing and recovering the block this way will change the UUID of the filesystem, and may also erase the volume label (I didn't check that).
But after it gets done the filesystem type will be changed to ext3 (mine was, even though it was mounted as ext4).
There might be some other, more precise ways to do this...
Well I have had LVM working on two 500Gig drives with no problems until the last update. Weird! Thanks for the great reply!
Last edited by Randallwright; 29th March 2011 at 04:25 AM. Reason: text changes
I am almost ready to format, re-install and turn off all automatic updates. It seems like every time I try to do an update or try uninstalling software, it gets all messed up.![]()
That was one of the reasons I quit using LVM.
It just seemed to be adding problems and not fixing them. Since I wasn't trying to use multiple disks in any format (raid0/1/5) LVM didn't seem to be necessary.
Fedora 13 had no problem with reusing the filesystem (I only had two disks, sda adn sdb), so I eliminated LVM. When I added a third drive, disk identities got goobered. The BIOS had them as 1/2/3, but Linux had them 1/3/2 for some reason. So I went to mounting by UUID with Fedora 13. Worked fine. Fedora 14 though .... the reused LVM partition (which had my /home filesystem) quit mounting, claiming something about invalid format. It would mount fine manually, just not during boot. The error messages always pointed to an "LVM2_member" partition.. but I was not using LVM.
The above fix corrected that issue, though after fixing it I realized that the only field introducing the error contained "LVM2_member" which I could see in the dump. If I had thought about it for another hour I would have thought to use the offset into the block (512 + offset) and just zeroed the 11 bytes containing LVM2_member.
I think that would have removed the mount error, though it is still possible that it could have complained about not matching the filesystem. It is now identified as a type=ext3, but I created it as ext4.
Now I realize ext3 and ext4 are NEARLY identical. And it is possible that I haven't caused files with extents yet (everything is relatively small files
or files I created when it was an ext3 filesystem), so the fsck using a recovery superblock could have caused that.