Veritas Storage Foundation 5.0 for Oracle RAC Configuration Guide Extracts for HP Serviceguard Storage Management Suite, Second Edition, May 2008

Planning SGeRAC Installation and Configuration
About CVM and CFS in an SGeRAC Environment
Chapter 2
30
About CVM and CFS in an SGeRAC Environment
Before installing SGeRAC, you can review concepts on CVM and CFS to better
understand the overall setup and plan your SGeRAC configuration.
• “About CVM” on page 30
• “About CFS” on page 31
• “Coordinating CVM and CFS Configurations” on page 32
• “About Shared Disk Groups” on page 32
• “About Raw Volumes Versus CFS for Data Files” on page 34
About CVM
Review CVM configuration differences from VxVM and CVM recovery operations.
CVM Configuration Differences
CVM configuration differs from VxVM configuration in these areas:
• Configuration commands occur on the master node.
• Disk groups are created (could be private) and imported as shared disk groups.
• Disk groups are activated per node.
• Shared disk groups are automatically imported when CVM starts.
CVM Recovery
When a node leaves a cluster, it can leave some mirrors in an inconsistent state. The
membership change is communicated through GAB to the vxconfigd daemon, which
automatically calls the vxrecover utility with the -c option when necessary.
CVM supports both the FastResync option and dirty region logging (DRL) as optional
features to improve resynchronization performance. FastResync improves performance
when reorganizing volumes (moving, splitting, and joining disk groups). This is useful
when performing off-host processing. DRL speeds up resynchronization after a node
failure.
Special considerations exist when using the DRL in an SGeRAC environment. As in a
non-clustered environment, the DRL in clusters exists on a log subdisk in a mirrored
volume. The size of the DRL in clusters is typically larger than in non-clustered systems.
The log size depends on the volume size and the number of nodes. The vxassist
command automatically imports a sufficiently large DRL.
You can reimport a private disk group as a shared disk group but the DRL for any
mirrored volume in the disk group is probably too small to accommodate maps for all the
cluster nodes. Adding nodes to the cluster can also result in too small a log size. In this
situation, VxVM marks the log invalid and performs full volume recovery instead of
using DRL.