Filesystem still has errors

Filesystem still has errors

[Raimund andersen]

Hi,

I'm using ext4 as a root fs on my 2.6.28-gentoo-r1 (x86_64), and after a few
locks due to X misconfiguration, which forced me to hard-reset, I sometimes get
the 'no space left on device' error message when trying to create new files. But
that's not true:

#>df
Filesystem           1K-blocks      Used Available Use% Mounted on
rootfs               201585632  16234952 175110680   9% /
/dev/root            201585632  16234952 175110680   9% /
(...)

#>df -i
Filesystem            Inodes   IUsed   IFree IUse% Mounted on
rootfs               12804096  520672 12283424    5% /
/dev/root            12804096  520672 12283424    5% /
(...)

#>fsck.ext4 -fnv /dev/sda9
e2fsck 1.41.4 (27-Jan-2009)
Warning!  /dev/sda9 is mounted.
Warning: skipping journal recovery because doing a read-only filesystem check.
Pass 1: Checking inodes, blocks, and sizes
Deleted inode 447291 has zero dtime.  Fix? no

Pass 2: Checking directory structure
Pass 3: Checking directory connectivity
Pass 4: Checking reference counts
Pass 5: Checking group summary information
Inode bitmap differences:  -447291
Fix? no


/dev/sda9: ********** WARNING: Filesystem still has errors **********


  520672 inodes used (4.07%)
     379 non-contiguous files (0.1%)
     258 non-contiguous directories (0.0%)
         # of inodes with ind/dind/tind blocks: 0/0/0
         Extent depth histogram: 512092/62
 4862330 blocks used (9.50%)
       0 bad blocks
       2 large files

  468666 regular files
   42140 directories
     157 character device files
      31 block device files
       2 fifos
    5149 links
    9643 symbolic links (8294 fast symbolic links)
      23 sockets
--------
  525811 files


One reboot earlier, the messages where similar:

#> fsck.ext4 -nv /dev/sda9
e2fsck 1.41.4 (27-Jan-2009)
Warning!  /dev/sda9 is mounted.
Warning: skipping journal recovery because doing a read-only filesystem check.
/dev/sda9 contains a file system with errors, check forced.
Pass 1: Checking inodes, blocks, and sizes
Deleted inode 447291 has zero dtime.  Fix? no

Pass 2: Checking directory structure
Pass 3: Checking directory connectivity
Pass 4: Checking reference counts
Pass 5: Checking group summary information
Free blocks count wrong (46338006, counted=46337933).
Fix? no

Inode bitmap differences:  -447291
Fix? no

Free inodes count wrong (12283475, counted=12283437).
Fix? no


/dev/sda9: ********** WARNING: Filesystem still has errors **********


  520621 inodes used (4.07%)
     378 non-contiguous files (0.1%)
     258 non-contiguous directories (0.0%)
         # of inodes with ind/dind/tind blocks: 0/0/0
         Extent depth histogram: 512079/62
 4861994 blocks used (9.50%)
       0 bad blocks
       2 large files
                                             
  468655 regular files
   42138 directories
     157 character device files
      31 block device files
       2 fifos
    5149 links
    9643 symbolic links (8294 fast symbolic links)
      23 sockets
--------
  525798 files
                                              

of course, I hesitate to call e2fsck.ext4 without -n on my mounted root-fs.
But when being invoked from an outside installed linux, no errors are detected:

#> fsck.ext4 -fnv /dev/sdb9
e2fsck 1.41.4 (27-Jan-2009)
Pass 1: Checking inodes, blocks, and sizes
Pass 2: Checking directory structure
Pass 3: Checking directory connectivity
Pass 4: Checking reference counts
Pass 5: Checking group summary information

  520634 inodes used (4.07%)
     379 non-contiguous files (0.1%)
     258 non-contiguous directories (0.0%)
         # of inodes with ind/dind/tind blocks: 0/0/0
         Extent depth histogram: 512068/62
 4862059 blocks used (9.50%)
       0 bad blocks
       2 large files

  468644 regular files
   42138 directories
     157 character device files
      31 block device files
       2 fifos
    5149 links
    9642 symbolic links (8293 fast symbolic links)
      11 sockets
--------
  525774 files


The same is true with putting a '/forcefsck'; the filesystem is checked and no
error message appears.
But when invoking fsck manually after full bootup, fs is still corrupted. 

So what should I do?

Any help is greatly apreciated.

Raimund

Re: Filesystem still has errors

[Eric Sandeen]

Raimund andersen wrote:
> Hi,
> 
> I'm using ext4 as a root fs on my 2.6.28-gentoo-r1 (x86_64), and after a few
> locks due to X misconfiguration, which forced me to hard-reset, I sometimes get
> the 'no space left on device' error message when trying to create new files. But
> that's not true:
> 
> #>df
> Filesystem           1K-blocks      Used Available Use% Mounted on
> rootfs               201585632  16234952 175110680   9% /
> /dev/root            201585632  16234952 175110680   9% /
> (...)
> 
> #>df -i
> Filesystem            Inodes   IUsed   IFree IUse% Mounted on
> rootfs               12804096  520672 12283424    5% /
> /dev/root            12804096  520672 12283424    5% /
> (...)
> 

snip ...

> The same is true with putting a '/forcefsck'; the filesystem is checked and no
> error message appears.
> But when invoking fsck manually after full bootup, fs is still corrupted. 
> 
> So what should I do?

You should expect that checking a live filesystem will produce errors,
actually.  So don't worry about that.  The filesystem is not quiesced,
and it's not consistent.  That is to be expected, and it's why you see
the different behavior w/ forcefsck vs. e2fsck -n on the live fs.

As for the spurious ENOSPC, I'm not certain.  Do you still get the error?

> Any help is greatly apreciated.
> 
> Raimund

Re: Filesystem still has errors

[Raimund Andersen]

Eric Sandeen <sandeen <at> redhat.com> writes:

> 
> You should expect that checking a live filesystem will produce errors,
> actually.  So don't worry about that.  The filesystem is not quiesced,
> and it's not consistent.  That is to be expected, and it's why you see
> the different behavior w/ forcefsck vs. e2fsck -n on the live fs.
> 
> As for the spurious ENOSPC, I'm not certain.  Do you still get the error?

Thanks for your quick reply. No, at the moment, I don't get it. Yesterday, it
occured once and a reboot let it vanish, and I got it some days before. Also
then, a reboot let it disappear. So could it be of help if I collect any
information in the case it might reappear? What exactly should I do then?

Thank you, once again.

rgds Raimund

Re: Filesystem still has errors

[Eric Sandeen]

Raimund Andersen wrote:
> Eric Sandeen <sandeen <at> redhat.com> writes:
> 
>> You should expect that checking a live filesystem will produce errors,
>> actually.  So don't worry about that.  The filesystem is not quiesced,
>> and it's not consistent.  That is to be expected, and it's why you see
>> the different behavior w/ forcefsck vs. e2fsck -n on the live fs.
>>
>> As for the spurious ENOSPC, I'm not certain.  Do you still get the error?
> 
> Thanks for your quick reply. No, at the moment, I don't get it. Yesterday, it
> occured once and a reboot let it vanish, and I got it some days before. Also
> then, a reboot let it disappear. So could it be of help if I collect any
> information in the case it might reappear? What exactly should I do then?

If you can, I'd test a 2.6.29-rc4 kernel, see if the very latest code is
behaving any better...

-Eric

> Thank you, once again.
> 
> rgds Raimund

Re: Filesystem still has errors

[Raimund Andersen]

Eric Sandeen <sandeen <at> redhat.com> writes:

> If you can, I'd test a 2.6.29-rc4 kernel, see if the very latest code is
> behaving any better...
> 
...ok, I'll keep an eye on it ;-)

Raimund