 | |||||||||||
| |||||||||||
 |
|||||||||||
| |
| ||||||||||||||||
| ShaoLin HA Cluster 1.0 Installation Manual and Operations Guide | ||
|---|---|---|
| Prev | Chapter 10. Cluster Administration | Next |
You can use Linux LVM, software RAID or SAN for storage management systems together with ShaoLin HA Cluster. It is important you can make changes at the active server and let the standby server to pick up those changes in case of system failover. Most of the software level storage configurations can be automatically synchronize between the clusters, however there are cases that you will need to restart your standby server to make the standby server to pick up those changes after reconfiguration. This section illustrate how it works for common storage management tools and storage subsystems. If you are looking for information regarding adding or removing storage device to/from your system, please see Section 10.4 and Section 10.4.2.
If any changes to partition table is made at the active server after the standby server boots, it is necessary to reboot the standby server to take effect. Fail to do so will result in data corruption due to incorrect partition table recognition by the standby server after failover.
If you are using shared SCSI storage and you have added new disks to the disk array by hot plugging disk drives into the system, your standby server will not able to detect those new drives after system failover. You will require to reboot the standby server to take effect. If you are using a disk array enclosure with built-in hardware RAID controller, provided your disk enclosure able to handle hot-plug and is transparent to the operating systems, you may hot-plug disk drives into the disk array without rebooting any machines.
If you use software RAID (MD) or Linux Logical Volume Manager (LVM), it is safe to alter configurations and the settings at the active server without the need to reboot the standby server, provided you don't make any changes to the physical partitions.
SAN virtualized storage architecture and some storage enclosures that has a built-in disk array controller allow you to expand the size of disk array and create snapshot partitions on-line. Usually, it works independent of the operating system. For these changes, the physical disk size is changed without notifying the operating system or disk drivers. One way to pick up these changes by the operating system is to rescan the SCSI bus. The best way to pick up changes without causing too much downtime is first reboot the standby server, and then manually trigger failover using Cluster Manager. This will cause the standby server to failover the system with newly detected volumes. Then you can make necessary settings on the system on-line.
