5.5.1 Shared disk configuration
A shared disk is controlled for each volume group.
To control a shared disk for each volume group, you must connect the shared disk from all the hosts that use the shared disk and control it by using LVM. Place the data that is used by each server in a separate group and provide an independent volume group to each server. If you use a resource server, you can share a single shared disk among multiple servers.
- Preventing a single point of failure
-
HA Monitor does not directly handle disk failures or failures on I/O paths such as disk adapters. Use both of the following methods to create a configuration that prevents a shared disk from becoming a single point of failure:
-
Data redundancy using RAID
Achieve data redundancy by using RAID devices, such as Hitachi disk array system.
-
I/O path redundancy
Achieve I/O path redundancy by using disk path switching software, such as HDLM.
-
If you use SCSI reservation for shared disk, you need a shared disk. LVM mirroring and Linux standard software RAID (md) are not supported.
- Note
-
In the event that SCSI reservation for shared disk is used, HA Monitor supports HDLM, DMMP, EMPD, and HFC-PCM as multipath software. For details about setting up multipath software, see 6.4.1 Setting up multipath software.