Red Hat GFS 6.0 -, Руководство администратора

Страница: 142 / 192

128 Chapter 12. Using GFS
init.d
Scripts
12.2. GFS
init.d
Scripts Use
The following example procedure demonstrates using the GFS
init.d
scripts to start GFS:
1. Install GFS on each node.
2. Load the
pool
module:
Note
If you need to specify a persistent major number, edit
/etc/modules.conf
before loading
pool.o
. Refer to Section 3.1.2 Specifying a Persistent Major Number
#
modprobe pool
or
#
service pool start
3. Create pool labels.
4. Write the labels to disk.
5. Load the pools. You can use the
init.d
script to reload or rediscover pool labels as follows:
#
service pool restart
You have the option of specifying in a configuration file
/etc/sysconfig/gfs
the pools that
you want assembled. If no pools are specified, then the
pool
script scans all the devices and
loads any pool that it finds.
To specify the pools on which to operate, the variable
POOLS
must be set in
/etc/sysconfig/gfs
. You can define multiple pools by separating the pool names with a
space, as shown in the following example:
POOLS="trin.cca trin1.gfs"
6. Create the CCS archive.
7. Write the CCS archive to disk or to a file.
8. Modify
/etc/sysconfig/gfs
to specify the location of the CCS archive by defining the vari-
able
CCS_ARCHIVE
in
/etc/sysconfig/gfs
. For example:
CCS_ARCHIVE="/dev/pool/trin.cca"
9. Start
ccsd
as follows:
#
service ccsd start
If
CCS_ARCHIVE
is not defined in
/etc/sysconfig/gfs
,
pool_tool
is used to scan for
assembled pools that have CCS archives. If a single archive is found, then that device is auto-
matically used.
10. Start
lock_gulmd
as follows:
#
service lock_gulmd start
No additional configuration is required.
ccsd
needs to be running.
11. Create GFS file systems using the
gfs_mkfs
command.