The DiskGroup agent has the following notes:
The high availability fire drill detects discrepancies between the VCS configuration and the underlying infrastructure on a node. These discrepancies might prevent a service group from going online on a specific node.
For DiskGroup resources, the high availability fire drill checks for:
For more information about using the high availability fire drill see the Veritas Cluster Server User's Guide.
When you use a volume set, set StartVolumes and StopVolumes attributes of the DiskGroup resource that contains a volume set to 1. If a file system is created on the volume set, use a Mount resource to mount the volume set.
See the UNIX Mount agent for more information.
VCS requires that the noautoimport flag of an imported disk group be explicitly set to true. This value enables VCS to control the importation and deportation of disk groups as needed when bringing disk groups online and taking them offline.
To check the status of the noautoimport flag for an imported disk group
# vxprint -ldisk_group
| grep noautoimport
If the output from this command is blank, the noautoimport flag is set to false and VCS lacks the necessary control.
and VxVM versions 4.1 and 5.0 for Solaris
The Monitor function changes the value of the VxVM noautoimport flag from off to on. It changes the value instead of taking the service group offline. This action allows VCS to maintain control of importing the disk group.
The following command changes the autoimport flag to false:
# vxdg -g disk_group set autoimport=no
For VxVM version 4.0
When you enable a disk group that is configured as a DiskGroup resource that does not have the noautoimport flag set to true, VCS forcibly deports the disk group. This forcible deportation may disrupt applications running on the disk group.
To explicitly set the noautoimport flag to true, deport the disk group and import it with the -t option as follows:
To deport the disk group, enter:
# vxdg deport disk_group
To import the disk group, specifying the noautoimport flag be set to true to ensure that the disk group is not automatically imported, enter:
# vxdg -t import disk_group
Linux and Solaris: Most Fiber Channel (FC) drivers have a configurable parameter called "failover." This configurable parameter is in the FC driver's configuration file. This parameter is the number of seconds that the driver waits before it transitions a disk target from offline to failed. After the state becomes failed, the driver flushes all pending fiber channel commands back to the application with an error code. Symantec recommends that you use a non-zero value that is smaller than any of the MonitorTimeout values of the Disk Group resources. Use this value to avoid excessive waits for monitor timeouts.
AIX: Refer to the Fiber Channel adapter's configuration guide for further information.