HP StorageWorks Scalable File Share User Manual
Page 257
Troubleshooting file system performance
9–33
2.
Determine whether the performance
anomaly can be reproduced.
1.
Rerun the client application.
2.
Ensure that the timing measurements are being made
correctly, by synchronizing the time on the client nodes and
on the servers in the HP SFS system. Use the NTP functionality
to ensure accurate synchronization. Refer to Chapter 6 and
Chapter 7 of the HP StorageWorks Scalable File Share
System Installation and Upgrade Guide and Chapter 2 and
Chapter 3 of the HP StorageWorks Scalable File Share Client
Installation and User Guide for more information.
The
IOR
test that is run as part of the client performance
diagnostics (see Task number 3 in this table) indicates the
time deviation between the clients on which it runs.
3.
Look out for intermittent results, and determine whether they
are being caused by one of the following problems:
• The file system is mounted by other clients, with
applications actively accessing the file system.
• The interconnect bandwidth is being used by other
applications at the same time as it is being used for
communications with the servers in the HP SFS system.
3.
Determine the performance of the OST
services.
Run performance diagnostics on the client nodes, as described
in Section 6.1.9. Run the diagnostics in the sequence shown in
4.
Ensure that Lustre is configured
correctly.
1.
If Lustre debugging is enabled, disable it (see Section 9.9).
2.
Verify that the file is being distributed across all OST services
3.
Ensure that the server is not running in a failed-over
configuration (see Section 9.27.3 and Section 9.27.4).
5.
Determine whether errors are being
recorded.
1.
Examine the logs on the client node.
2.
Examine the server logs (see Section 9.27.5).
3.
Examine the storage subsystems for errors (see Section 9.27.6
or Section 9.27.7, depending on the storage used).
Table 9-1
Performance troubleshooting tasks
Task
Process