[Lustre-discuss] LAST_ID issue on lustre 1.8.7

Andrus, Brian Contractor bdandrus at nps.edu
Thu Jun 21 15:06:42 PDT 2012


All,

We seem to be having a bad failure here.
Here is some info from lctl and from looking at the lov_objid for our /work filesystem:
================================
[root at nas-0-1 ~]# mount /mnt/lustre/work/mdt
[root at nas-0-1 ~]# lctl dl
  0 UP mgs MGS MGS 5
  1 UP mgc MGC10.100.255.89 at o2ib 87c71980-91c9-51ae-bf2c-37113d7a06e7 5
  2 UP mdt MDS MDS_uuid 3
  3 UP lov work-mdtlov work-mdtlov_UUID 4
  4 UP mds work-MDT0000 work-MDT0000_UUID 3
  5 UP osc work-OST0000-osc work-mdtlov_UUID 5
  6 UP osc work-OST0003-osc work-mdtlov_UUID 5
  7 UP osc work-OST0005-osc work-mdtlov_UUID 5
  8 UP osc work-OST0002-osc work-mdtlov_UUID 5
  9 UP osc work-OST0009-osc work-mdtlov_UUID 5
 10 UP osc work-OST0001-osc work-mdtlov_UUID 5
 11 UP osc work-OST0007-osc work-mdtlov_UUID 5
 12 UP osc work-OST0006-osc work-mdtlov_UUID 5
 13 UP osc work-OST0008-osc work-mdtlov_UUID 5
 14 UP osc work-OST0004-osc work-mdtlov_UUID 5
[root at nas-0-1 ~]# !umount
umount /mnt/lustre/work/mdt
[root at nas-0-1 ~]# mount -t ldiskfs /dev/VG_hamming/work_mdt /mnt/lustre/work/mdt
[root at nas-0-1 ~]# od -Ax -td8 /mnt/lustre/work/mdt/lov_objid
000000             77090836             87265628
000010            104094987             84525048
000020             70980680             36265791
000030             36262683             36068567
000040             35917014             26909787
000050
[root at nas-0-1 ~]#
==============================================

What concerns me is that we have 10 OSTs for the filesystem, but there appear to only be 6 showing up in the lov_objid...
Am I reading this correctly? If so how do I recover those 'missing' OSTs.

Brian



More information about the lustre-discuss mailing list