[DRBD-user] Filesystem Inconsisteny

Stay, Stefan (EXT) stefan.stay.ext at siemens.com
Thu Mar 9 09:31:39 CET 2006

Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.


Hello list,

we use an auto-install-mechanism, to install our software load on our
systems. One of
the tasks, that will be accomplished by this routine is to create
filesystems on partitions,
mount them and install the software load.

The partitions mentioned above will be threaded, respectively used as
DRBD-Devices later,
after the first boot of the systems.
Therefore the installation-routine will create filesystems, with respect
of the metadata-area,
so the filesystem-size (in fs-blocks) that is given by the
"mkfs"-command is reduced about 128MB.

After the creation of the filesystem, it is mounted and the software
load is installed. Just after rebooting
after the installation, the partitions with the existent filesystems,
will be setup as drbd-devices. The metadata
is held internal.

Now the question:

We inspected that the filesystem has errors after the
auto-install-routine had cleanly unmount the filesystems, which
are prepared for the use as drbd-devices.

We could give you the following info:

Kernel: 2.6.10_mvlcge40
DRBD: 0.7.10
Filesystem: ext2 (will later be migrated to ext3)

Actions which were performed:

The software-load is installed on both, primary and secondary node. Only
the drbd-content
will installed and extracted to the primary node and later after the
first boot of both nodes,
mirrored using drbd.

1. mkfs -t ext2 -b 2048 /dev/discs/disc0/part11 6078976
	(the filesystems original size is 12000MB so only 11873MB will
be use for the filesystem)

2. mount the lower-level device on temp. mp 

3. extract the software-load to it (4.3GB of diskless blades OS -
Repositorys)

4. unmount the lower-level device

5. booting both systems with their freshly installed software-load
(node1 will the Sync-Source)

6. poweroff node2 (secondary - synctarget)

7. unmount the drbd-devices on node1 and checking it for errors (e2fsck)
gave the following output:

size of /export (/dev/nbd/5)
/dev/nbd/5            11965818   4285744   7072178  38% /export

root at sb1-1:~# umount /export/
root at sb1-1:~# e2fsck /dev/nbd/5
e2fsck 1.35 (28-Feb-2004)
/dev/nbd/5 was not cleanly unmounted, check forced.
Pass 1: Checking inodes, blocks, and sizes
Inode 1085822, i_blocks is 528, should be 500.  Fix<y>? yes

Inode 1085821, i_blocks is 2588, should be 2572.  Fix<y>? yes

Inode 1139871, i_blocks is 2584, should be 2572.  Fix<y>? yes

Inode 1139872, i_blocks is 520, should be 500.  Fix<y>? yes

Pass 2: Checking directory structure
Pass 3: Checking directory connectivity
Pass 4: Checking reference counts
Pass 5: Checking group summary information

/dev/nbd/5: ***** FILE SYSTEM WAS MODIFIED *****
/dev/nbd/5: 111035/1523712 files (0.6% non-contiguous), 2238939/6078976
blocks


Who can tell me, why the filesystem has this inconsistency. The
filesystem
was never "not cleanly unmounted". Thanks...

	Stefan 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linbit.com/pipermail/drbd-user/attachments/20060309/44189a94/attachment.htm>


More information about the drbd-user mailing list