IBM SG24-5131-00 사용자 설명서

다운로드
페이지 240
126
 
IBM Certification Study Guide  AIX HACMP
To prevent problems with NFS file systems in an HACMP cluster, make sure 
that each shared volume group has the same major number on all nodes. The 
lvlstmajor
 command lists the free major numbers on a node. Use this 
command on each node to find a major number that is free on all cluster 
nodes, then, record that number in the Major Number field on the 
Shared 
Volume Group/File System (Non-Concurrent Access) worksheet in Appendix 
A, Planning Worksheets, of the 
HACMP for AIX, Version 4.3: Planning Guide, 
SC23-4277 for a non-concurrent access configuration. 
Alternatively, if you use the Task Guide to create your shared volume groups, 
it will make sure that the major number is the same on all nodes that will 
share it.
5.4.2  Exporting NFS File Systems
The default scripts provided with HACMP do not use the /etc/exports file. 
Instead, the default scripts provided call a 
cl_export_fs
 utility that uses the 
exportfs
 command with the 
-i 
flag and specifies the file system names stored 
in the HACMP ODM object class.
Therefore export options specified in the /etc/exports
 
file are ignored. 
However, export options may be specified by modifying the 
cl_export_fs
 
utility. Alternately, the /etc/exports file can be used as is typical in an NFS 
environment by simply removing the
 -i
 flag from the 
exportfs
 command in 
the cl_export_fs utility.
5.4.3  NFS Mounting
For HACMP and NFS to work together properly, you must be aware of the 
following mount issues:
5.4.3.1  Creating NFS Mount Points on Clients
A mount point is required in order to mount a file system with NFS. Mount 
points are required for NFS clients, not servers; however, you should be 
aware that a server can also be a client.
5.4.4  Cascading Takeover with Cross Mounted NFS File Systems 
This section describes how to set up cascading resource groups with cross 
mounted NFS file systems.
5.4.4.1  Server-to-Server NFS Cross Mounting
HACMP allows you to configure a cluster so that servers can NFS-mount 
each other’s file systems. The following figure shows an example: