NFS needs unique major numbers for vxio and vxspec on cluster nodes.
Updated: 03 Mar 2010 | 2 comments
Status:
Implemented
Most of the VCS admins are aware that if they have to run NFS HA, they have to ensure that respective major number for vxio and vxspec is unique across all the cluster nodes. Currently, VCS install script does not check for any discrepancy on this. The experienced [and/or trained] VCS admins know that this is "must" check before building the cluster, many relatively new VCS admins fail to check this before hand and face problems.
How about the VCS install script checking this and flagging if the respective major number for vxio and vxspec is not unique on all nodes in a cluster?
idea Filed Under:
Comments
You can accomplish this with CNFS without being concerned
Hi Sunil
Two major enhancements are coming in the 5.1 version of CFS HA in November.
1 - The ability to do simultaneous NFS (which we call Clustered NFS or CNFS) off of each node in a cluster file system cluster.
This obviates the concern around major numbers and import issues in the event of node failure.
2 - The issue of conflicting major numbers should be resolved in general in dg import operations. This is particularly
for CVM where cluster import can lead to these problems.
Implemented in 5.1
Implemented in 5.1
Would you like to reply?
Login or Register to post your comment.