[LU-722] Create mdsdb exit with failure

• Type: Bug
• Resolution: Duplicate
• Priority: Major
• Affects Version/s: Lustre 1.8.6
• Labels: e2fsprogs
• Environment: RHEL 5.5 with kernel from Oracle Lustre 1.8.5 + e2fsprogs-1.41.90.wc3-0redhat
• Severity: 3

Epic

metadata server

• Assignee: WC Triage
• Reporter: Marian Hromiak

Description

Hello,

when we tried generate mdsdb from lustre 1.8.5 via e2fsprogs-1.41.90.wc3-0redhat we obtain following error:
e2fsck -nv --mdsdb /tmp/mdsdb /dev/vgl6mdt/lvol1

e2fsck 1.41.90.wc3 (28-May-2011)
device /dev/mapper/vgl6mdt-lvol1 mounted by lustre per /proc/fs/lustre/mds/l6-MDT0000/mntdev
Warning! /dev/vgl6mdt/lvol1 is mounted.
Warning: skipping journal recovery because doing a read-only filesystem check.
l6-MDTffff has been mounted 24 times without being checked, check forced.
Pass 1: Checking inodes, blocks, and sizes
Inode 11 creation time (Sun Feb 8 02:38:54 1970) invalid.
Clear? no

MDS: ost_idx 0 max_id 25693660
MDS: ost_idx 1 max_id 25622279
MDS: ost_idx 2 max_id 25965512
MDS: ost_idx 3 max_id 25795531
MDS: got 32 bytes = 4 entries in lov_objids
MDS: max_files = 3512383
MDS: num_osts = 4
mds info db file written
error: only handle v1/v3 LOV EAs, not 00000001
e2fsck: aborted

Could you help us, how we can dump mds database for distributed client check?

Issue Links

• Trackbacks: Lustre 1.8.x known issues tracker While testing against Lustre b18 branch, we would hit known bugs which were already reported in Lustre Bugzilla https://bugzilla.lustre.org/. In order to move away from relying on Bugzilla, we would create a JIRA