[Lustre-discuss] incorrect Target for OST

[Mohr Jr, Richard Frank (Rick Mohr)]

On Jul 18, 2014, at 7:51 AM, Gretchen Zwart <zwart at astro.umass.edu> wrote:

> I have 2 of 10 OSTs reporting incorrect Target (OST0000 rather than
> OST0006,OST0007). I tried unmounting MDT/MGS and all OSTs and "tunefs
> --writeconf" but "tunefs --print" still reports incorrect Target.
> Mounted as ldiskfs the OSTs show *incorrect* value in
> CONFIGS/mountdata but lustre-client log file looks as if it is
> reporting correct OST name for the OST server/OSTs in question (i.e.
> OST0006 shows up in  /xxxx/CONFIGS/mountdata/lustre-client).


The tunefs.lustre command has a "--index" option, but when I tried using it on a formatted OST, the command refused to modify the index.  However, when I added the --writeconf option as well, then the index could be changed:

#> tunefs.lustre --dryrun --writeconf --index=100 /dev/sdb
checking for existing Lustre data: found
Reading CONFIGS/mountdata

   Read previous values:
Target:     brize-OST0001
Index:      1
Lustre FS:  brize
Mount type: ldiskfs
Flags:      0x402
              (OST )
Persistent mount opts: errors=remount-ro,extents,mballoc
Parameters: mgsnode=172.16.3.102 at o2ib ost.quota_type=ug


   Permanent disk data:
Target:     brize=OST0064
Index:      100
Lustre FS:  brize
Mount type: ldiskfs
Flags:      0x502
              (OST writeconf )
Persistent mount opts: errors=remount-ro,extents,mballoc
Parameters: mgsnode=172.16.3.102 at o2ib ost.quota_type=ug

exiting before disk write.


Maybe you could use this to correct the index.  (WARNING: I have never tried this on a production file system so I don't know if there would be any negative side effects.)

-- 
Rick Mohr
Senior HPC System Administrator
National Institute for Computational Sciences
http://www.nics.tennessee.edu