HP StorageWorks Scalable File Share - Manual (Page 260)

Troubleshooting 9–36
9.27.2 Verifying file striping
If a file is not being striped across all OST services, the full bandwidth may not be available. This can occur
in the following circumstances:
• If the file was initially created with the
lfs setstripe
command to a specific striping configuration
• If a file inherited an incorrect default stripe setting from the containing directory
• If the file system default striping is incorrect
You can check the file striping configuration as follows:
• To check the striping for an individual file, enter the
lfs getstripe file_name
command
(where
file_name
is the name of a file on which read and write performance problems are being
experienced), as shown in the following example. Enter the command on the client node that is
experiencing the problems:
# lfs getstripe data.all
OBDS:
0: OST_south2_UUID
1: OST_south2_2_UUID
2: OST_south2_3_UUID
3: OST_south2_4_UUID
data.all
obdidx objid objid group
1 1 0x1 0
2 1 0x1 0
3 1 0x1 0
0 1 0x1 0
# lfs getstripe data.1
OBDS:
0: OST_south2_UUID
1: OST_south2_2_UUID
2: OST_south2_3_UUID
3: OST_south2_4_UUID
data.1
obdidx objid objid group
1 2 0x2 0
In this example, the performance on the
data.1
file will be poor, as the file is striped across only a
single OST service, whereas the
data.all
file is striped across four OST services.
• To check the file striping in a directory, enter the command shown in the following example on the
client node, where
/mnt/lustre/one-stripe
is the name of the directory containing the files
that are experiencing performance problems:
# lfs find -v /mnt/lustre/one-stripe
OBDS:
0: OST_south4_UUID
1: OST_south4_2_UUID
2: OST_south4_3_UUID
3: OST_south4_4_UUID
/mnt/lustre/one-stripe/
count: 1, size: 0, offset: -1
• To check the file system default striping, enter the
show filesystem
command as follows:
sfs> show filesystem
Name State Services
------------- -------------- ----------------------------------
scratch started mds1: running, ost[1-4]: running