HP StorageWorks Scalable File Share User Manual
Page 91
Creating a file system — EVA4000 storage
5–7
14 4 B ost 290 south[5-6] -
15 5 A ost 290 south[5-6] -
16 5 B ost 290 south[5-6] -
.
.
.
You must select preferred servers so that OST serving is balanced optimally. Both of the servers in a server
pair are capable of serving (that is, being the preferred server for) any of the LUNs on the arrays attached
to the server. In the example above, LUNs 8 through 11 are visible to (that is, presented to) the
south3
and
south4
servers. This means that either of these servers could be the preferred server for any of LUNs 8
through 11. However, it would not be optimal to prefer all of the LUNs to one of the servers, as this would
leave the second server unused. To help you to balance the distribution of LUNs to servers, the
create
filesystem
command shows the EVA4000 controller that is managing each LUN.
The HP SFS system is configured with multiple paths between a server and the EVA4000 array. Each host
has two HBA ports. Each HBA port is connected by a separate Fibre Channel fabric to both EVA4000
controllers. This means that each server has four paths to a given LUN.
However, when you create a LUN, you specify the preferred EVA4000 controller for the LUN (refer to
Chapter 5 of the HP StorageWorks Scalable File Share System Installation and Upgrade Guide for more
information). In the example above, LUNs 8 and 9 are being managed by different EVA4000 controllers.
This is normal; however, if LUNs 8 and 9 were both managed by the same EVA4000 controller, you would
need to investigate the configuration and make changes if necessary. Specifically, if LUNs 8 and 9 were
both preferred to the
south3
server, the
south3
server would access both LUNs through the same
controller port and the second EVA4000 controller would be idle. This would impact performance.
For examples of file system configurations, see Appendix C.
For information on how to order OST services in the file system to support optimal performance, see
File system optimization for I/O pattern
You can optimize I/O operations on the OST services in a file system by specifying the file I/O pattern that
you expect on the file system. The options are:
•
Predominantly large file streamed sequential access
•
Large numbers of smaller files
Select the scenario that best suits your needs. If you choose the second option (large numbers of small files),
the
mballoc
option is added to the list of options that can be specified for OST services (this list is presented
later).
Mount options
Underlying the OST and MDS services on the Lustre file systems, there are
ldiskfs
file systems. When an
OST or MDS service starts, it mounts the underlying
ldiskfs
file system so that it can access the data. You
can specify mount options on the file system services to determine what happens when the file systems are
mounted.
•
When creating a file system, accept the default options provided by the
create filesystem
command for OST services.
•
You can optionally set the
acl
and
user_xattrs
options for MDS services:
•
The
acl
option must be enabled if Access Control Lists (ACLs) are to be used when the file
system is mounted on client nodes.
•
The
user_xattrs
option must be enabled if extended attributes are to be used when the file
system is mounted.