[Lustre-discuss] Find all files on a specific OST

Andreas Dilger adilger at sun.com
Wed Jan 30 15:08:02 PST 2008


On Jan 30, 2008  20:48 +0100, Anselm Strauss wrote:
> Yes, although it's a patchless client running on a 2.6.22 kernel. I  
> also tried it on the failover server that was not mounting any lustre  
> devices while I mounted the filesystem from the first server. It runs  
> a Vanilla 2.6.18.8 kernel with lustre patches and lustre version  
> 1.6.4.2. I got the same error.
> 
> I also did an strace for the command on my patchless client:
> 
> ...
> open("/proc/fs/lustre/lov/scratch-clilov-ffff8100f64bd000/target_obd",  
> O_RDONLY) = 4
> fstat(4, {st_mode=S_IFREG|0444, st_size=0, ...}) = 0
> mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1,  
> 0) = 0x2b163b43a000
> read(4, "", 1024)                       = 0
> close(4)                                = 0
> munmap(0x2b163b43a000, 4096)            = 0
> write(2, "error: setup_obd_uuid: unknown o"..., 77error:  
> setup_obd_uuid: unknown obduuid: scratch-OST0001-osc-ffff8100f64bd000
> ) = 77
> ...
> 
> Should there be any content in /proc/fs/lustre/lov/.../target_obd,  
> mine is empty?

Yes, this is definitely the source of the problem.  I do recall there
were some changes in the 2.6.22 kernel related to /proc/sys (not
allowing proc entries to be created under /proc/sys) but this may have
been more extensive...

Of course the release notes for the 2.6.22 kernel explained this all
clearly :-).

Cheers, Andreas
--
Andreas Dilger
Sr. Staff Engineer, Lustre Group
Sun Microsystems of Canada, Inc.




More information about the lustre-discuss mailing list