7.2.3 Notes about maintaining shared resources
This subsection provides important information about connecting and disconnecting the shared resources when you perform maintenance on a shared disk after operation. For the specific operating methods, see the OS documentation.
- Important
-
Pay attention to the notes provided here when you manipulate shared resources. If you ignore these notes, the shared resources might become corrupted or cause hot standby errors.
- Organization of this subsection
(1) Before you manipulate shared resources
HA Monitor controls the connection and disconnection of shared resources. Therefore, you must note the following:
-
While the servers are running, do not manipulate the shared resources from the outside such as by using commands.
If contention occurs between a user operation and the HA Monitor processing, the HA Monitor will not be able to control the system correctly.
-
If you need to manipulate shared resources for purposes such as maintenance, be sure to first terminate the servers.
For details about HA Monitor's control of shared resources, see 4.3.4 List of shared resource statuses.
(2) Handling volume groups
If a volume group created on a shared disk is accessed for referencing and updating from both hosts at the same time, operation is not guaranteed. Therefore, you must note the following:
-
Before you start the servers, make sure that the volume groups are disconnected from both hosts.
When a server is started, HA Monitor connects the corresponding volume group for referencing and updating. If the volume group is already connected when the server is started, the volume group might be referenced and updated from both hosts, resulting in data corruption.
-
After a server is terminated, HA Monitor disconnects the corresponding volume group. If another job is to use the volume group, connect it by executing the vgchange command.
In this case, make sure that the volume group is not referenced or updated from both hosts.
The following shows the vgchange command's execution formats:
-
Connecting for referencing and updating
vgchange -a y path-name-of-volume-group
-
Disconnecting
vgchange -a n path-name-of-volume-group
(3) Handling file systems
This subsection provides important information about handling file systems.
-
Before you start a server, check that the file system is unmounted at both hosts.
-
Mounting of a file system might fail if a process is accessing the mount point (target directory for mounting) before HA Monitor mounts the file system.
Avoid such handling of a file system.
-
If the operator uses a file system for another job, make sure that the file system is not mounted from both hosts.
If a file system created on the shared disk is mounted from both hosts at the same time, the file system itself becomes corrupted.
For the timing of mounting and unmounting of file systems by HA Monitor, see 4.3.2 Managing file systems.
(4) Handling of a shared disk after OS migration
The handling of a shared disk is not the same in Linux as in HI-UX/WE2 and AIX. If you have migrated your OS from HI-UX/WE2 or AIX to Linux, you must check and, if necessary, revise the handling of a shared disk and use it carefully.
For example, in HI-UX/WE2 and AIX, a shared disk can be accessed after the active server has terminated. In Linux, a shared disk cannot be accessed after the active server has terminated. In order to access a shared disk in Linux for a purpose such as making a backup, you must first place the shared disk in accessible status.