View previous topic :: View next topic |
Author |
Message |
jbc28 Apprentice
Joined: 07 Jan 2003 Posts: 205 Location: Edinburgh
|
Posted: Wed Dec 17, 2003 2:09 am Post subject: |
|
|
Hi,
has this problem been fixed?
I'm getting 'filesystem can't be fixed' on filesystems which happily pass
a manual fsck. I'm using ext3 and right now 'emerge world' gives me nothing to do so presumably my init scripts are up to date.
Thanks for any help,
J |
|
Back to top |
|
|
Azarah Retired Dev
Joined: 20 Jun 2002 Posts: 81
|
Posted: Wed Dec 17, 2003 6:10 am Post subject: |
|
|
I think it was a problem with /etc/init.d/halt.sh and the remount-readonly
code. Please try baselayout-1.8.6.12-r2 and let me know. |
|
Back to top |
|
|
grzewho l33t
Joined: 31 Dec 2002 Posts: 626 Location: /home/g
|
Posted: Wed Dec 17, 2003 10:36 am Post subject: |
|
|
my baselayout is 1.8.6.12-r2 and I get the same error. everytime the disks are being checked the system needs to be rebooted, but manual fsck gives no errors _________________
Code: | USE="freedom -software_patents" emerge --deep --update world |
|
|
Back to top |
|
|
jbc28 Apprentice
Joined: 07 Jan 2003 Posts: 205 Location: Edinburgh
|
Posted: Wed Dec 17, 2003 1:32 pm Post subject: |
|
|
Thanks for your fast reply (gentoo community really is something special!). How can I test this now? Otherwise I'll have to wait ~25 mounts for it to check again. Once it checks and works/doesn't I'll keep you posted.
Thanks for your help,
J |
|
Back to top |
|
|
dak1001 n00b
Joined: 16 Dec 2003 Posts: 9
|
Posted: Wed Dec 17, 2003 5:02 pm Post subject: sck of fsck? |
|
|
this is not a gentoo specific issue. more than anything, it is a failsafe. from an e2fsck standpoint:
e2fsck is prompted to check when the set mount count (-c #, i.e., -c 100, check after 100 mounts) is met e2fsck runs a check if the previous reboot/shutdown was not clean
you can alter the mount count using the '-c' switch with tune2fs,
Code: | tune2fs -c 100 /dev/sde3 |
you also have the option to set the current count,
Code: | tune2fs -C 100 /dev/sde3 | thus saying to the system that this partition has been mounted 100 times now, the next system restart we will do a general e2fsck.
now then, why the matter of the general e2fsck running and saying "cannot fix errors...enter password to repair...blah blah" is a failsafe:
e2fsck can perform systems repairs such as updating/correcting/repairing inodes e2fsck can perform a mere informative check and not actually do any repairs (the '-n' switch, i.e., e2fsck -n /dev/hda) e2fsck can also update/correct/repair the partition/filesystems superblock information (ever have one of those messages that basically say, "i don't know what the heck filesystem this is, your superblock is screwed, and so are you").
so, and this is not specific to gentoo, redhat, suse, slackware, or any other distro, from a system standpoint, we'd like to run an e2fsck and correct any "easily" corrected errors. if we cannot do that, then we should prompt the administrator to do a thorough check and alternatively, correct any errors.
the thing is, fixing inodes and such can cause the loss of data. this is why the system stops and says it cannot fix the problem.
if you do a bad shutdown/reboot, you may lose whatever data was in the buffers and not written to disk. but, some partial changes will likely have been written to the disk - so upon the next boot up, the system says, "hey, this bit here doesn't seem to match that over there as it should," i.e., inode is 852224 when it should be 2435, do you want to fix?
if you are confident that you do not need any of this data connected to a bad inode, go ahead and fix it.
otherwise, say no and use something like debugfs to try and rescue/retrieve whatever data you can.
also, make note of those lost+found directories...because after such an incident, the missing files/data/changes/etc. may be there.
hope this makes sense and helps. |
|
Back to top |
|
|
Azarah Retired Dev
Joined: 20 Jun 2002 Posts: 81
|
Posted: Wed Dec 17, 2003 5:32 pm Post subject: |
|
|
Ditto on what dak1001 said - the issue is the need to reboot at
every time.
For those with this issue - what fs again? I know we had some issues
with reiserfsck .... |
|
Back to top |
|
|
grzewho l33t
Joined: 31 Dec 2002 Posts: 626 Location: /home/g
|
Posted: Wed Dec 17, 2003 6:59 pm Post subject: |
|
|
ext3 _________________
Code: | USE="freedom -software_patents" emerge --deep --update world |
|
|
Back to top |
|
|
Azarah Retired Dev
Joined: 20 Jun 2002 Posts: 81
|
Posted: Thu Dec 18, 2003 7:20 pm Post subject: |
|
|
Tried a different kernel maybe? I use ext3, and no issue :/ |
|
Back to top |
|
|
jbc28 Apprentice
Joined: 07 Jan 2003 Posts: 205 Location: Edinburgh
|
Posted: Fri Dec 19, 2003 12:02 am Post subject: |
|
|
Hi,
Thanks for all your help. I've tried to reproduce the error but with the new base layout you suggested it seems to no longer be a problem (What have you changed?). Glad to have seemingly fixed this.
For the record I am and have been using gentoo-sources-2.4.20-r9 |
|
Back to top |
|
|
Plain-old-Jeb n00b
Joined: 19 Dec 2003 Posts: 51 Location: My House
|
Posted: Fri Dec 19, 2003 1:53 am Post subject: |
|
|
Quote: | For those with this issue - what fs again? I know we had some issues
with reiserfsck .... |
I had this problem using reiserfs, I decided to switch all my partitions to JFS and doing that found I had an identical problem (Stayed with JFS though it's very nice ).
I run fsck.jfs -f of the livecd on all my filesystems and they are all passed as clean. |
|
Back to top |
|
|
shwag n00b
Joined: 24 Sep 2003 Posts: 24 Location: Los Angeles, CA
|
Posted: Thu Jan 08, 2004 9:37 am Post subject: update |
|
|
Check the bug report listed in the earlier posts. It clearly shows the problem and the solution. The currently gentoo packages do not entirely contain the fix because it was only fixed in one of the two places needed. I have gotten the bug reopened, and hopefully it will be remedied soon since the fix is a simple patch to /etc/init.d/checkfs. |
|
Back to top |
|
|
|