#926130 e2fsck returns 1 which makes e2scrub fails

Package:
e2fsprogs
Source:
e2fsprogs
Description:
ext2/ext3/ext4 file system utilities
Submitter:
Laurent Bigonville
Date:
2021-02-14 16:21:08 UTC
Severity:
important
Tags:
#926130#5
Date:
2019-03-31 19:37:56 UTC
From:
To:
Hi,

For one of my LV, e2scrub consistantly fails.

Debugging this it seems that e2fsck returns 1 but it's not showing any
messages about reparing something.

According to the manpage, 1 means:
1    - File system errors corrected

If I'm umounting the FS and then run e2scrub, it returns 0

Any idea how to debug this further?

Regards,
Laurent Bigonville
#926130#10
Date:
2019-03-31 21:10:13 UTC
From:
To:
Can you send me (or attach to the bug) the output of running "e2scrub
<device>"?

Thanks,

					- Ted
#926130#15
Date:
2019-04-01 07:20:45 UTC
From:
To:
Le 31/03/19 à 23:10, Theodore Ts'o a écrit :
bigon@fornost:~$ sudo LC_ALL=C e2scrub /var/lib/sbuild/
   Logical volume "sbuildbuild.e2scrub" created.
e2fsck 1.45.0 (6-Mar-2019)
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
/dev/fornost/sbuildbuild.e2scrub: 13/983040 files (0.0% non-contiguous),
68098/3932160 blocks
/var/lib/sbuild/: Scrub FAILED due to corruption!  Unmount and run
e2fsck -y.
tune2fs 1.45.0 (6-Mar-2019)
Setting filesystem error flag to force fsck.
   Logical volume "sbuildbuild.e2scrub" successfully removed
#926130#20
Date:
2019-05-14 01:13:06 UTC
From:
To:
Hi.  I haven't forgotten about your bug report.  The problem has been
I can't find any explanation for how e2fsck could have returned an
exit status of 1 given the output shown below.

So... if you could do the following, I would greatly appreciate it.

1)  Confirm that this is still reproducing for you with e2fsprogs 1.45.0.

2) I've just released e2fsprogs 1.45.1 to Debian unstable.  If you
could try installing e2fsprogs, and see if you can still reproduce the
problem with e2fsprogs 1.45.1.

3) If you can, please create an /etc/e2fsck.conf containing the
following:

[options]
	max_count_problems = 2
	log_dir = /tmp
	problem_log_filename = e2fsck-problems-%N.%h.INFO.%D-%T
	log_filename = e2fsck-%N.%h.INFO.%D-%T
	log_dir_wait = true

4) The rerun the e2scrub command and send me the output of
/tmp/e2fsck-problems-*INFO* and /tmp/e2fsck-*INFO* files that will be
created as a result of running e2fsck.

Many thanks, and my apologies for not getting back to you sooner.

     	     	    	      	      	      - Ted
#926130#25
Date:
2019-05-14 07:42:58 UTC
From:
To:
Le 14/05/19 à 03:13, Theodore Ts'o a écrit :

Hello,

I also have troubles to understand what's happening

I can still reproduce that with 1.45.0 and 1.45.1

I've attached the requested logs. I don't see anything weird here.

I initially thought that this was caused by the fact that the VG was
split between a SSD and a regular HD, but now everything is on (2
different) SSD.

I've other LV's on the VG and they are not failing. FTR, the
"sbuildsbuild" LV has been created "2018-12-16 18:30:10 +0100" and the
docker one "2019-04-22 13:49:03 +0200". The other ones are way older,
like 2012, 2014 and 2017. So maybe something has changed in between?

$ mount |grep ext
/dev/mapper/fornost-root on / type ext4 (rw,relatime,seclabel,discard,errors=remount-ro)
/dev/sda3 on /boot type ext2 (rw,relatime,seclabel,block_validity,barrier,user_xattr,acl)
/dev/mapper/fornost-home on /home type ext4 (rw,relatime,seclabel,discard,stripe=22883)
/dev/mapper/fornost-acng--cache on /var/cache/apt-cacher-ng type ext4 (rw,relatime,seclabel,discard)
/dev/mapper/fornost-docker on /var/lib/docker type ext4 (rw,relatime,seclabel,block_validity,discard,delalloc,barrier,user_xattr,acl)
/dev/mapper/fornost-sbuildbuild on /var/lib/sbuild type ext4 (rw,relatime,seclabel,block_validity,discard,delalloc,barrier,user_xattr,acl)

I see differences in the options, but in fstab, they all shows
"defaults,discard"

Thanks to you!

Laurent Bigonville
#926130#30
Date:
2019-05-25 22:54:57 UTC
From:
To:
Yeah, I'm not seeing anything in the logs either.  :-(

Would you be willing to send me a compressed e2image file?  The
commands to generate are:

	e2image -Q /dev/fornost/docker docker.qcow2
	xz docker.qcow2

.. and then send me the docker.qcow2.xz file.  It will include only
the metadata blocks of the file system.  See the e2image man page for
more details.  This *will* include filenames in the directory blocks,
so if that makes you uncomfortable, you can use the -S option --- but
if it's all the same to you, it would be more convenient if you could
send it to me w/o using -S option.  The file will hopefully be small
enough that you can just e-mail it to me, or put it someplace where I
can download it (perhaps encrypted using GPG).

Thanks!!

					- Ted
#926130#35
Date:
2021-02-14 16:19:48 UTC
From:
To:
tag 926130 +moreinfo
thanks

Hi, it's been a while.  Are you seeing e2scrub problems, especially
with newer versions of e2fsprogs?  If you are, would you be willing to
send me a compressed e2image file?

Thanks!

						- Ted