core tools of ClusterPack, including PCC ClusterWare Pro™ and the HP Systems Insight
Manager. Node groups are collections of nodes that are subsets of the entire node membership
of the compute cluster. They may have overlapping memberships such that a single node may
be a member of more than one group. The node grouping mechanism allows flexible
partitioning of a compute cluster into logical collections that match their use model.
The first form of this command allows the user to add node groups to a compute cluster. The
initial definition of the node group can be specified as a list of individual nodes and/or other
groups. When a previously existing group is used in the formation of a new group, all
members of the pre-existing group are added to the new group.
The second form allows the user to remove a node group or nodes from a node group. When
removing nodes from a group, the nodes to be removed can be specified in terms of a list of
individual nodes and/or other groups. When a previously existing group is specified, all
members of that group are removed from the group being modified.
The third form allows the information regarding one or more node groups to be provided in a
file.
The last form lists all the node groups in the compute cluster.
clgroup is used as follows:
% clgroup -a groupname node1|group [node2|group ...]
% clgroup -r groupname [node1|group node2|group ...]
% clgroup -l groupname
% clgroup
Examples
The following command creates a node group "group1" containing compute cluster nodes
"nodea", "nodeb", and "nodec".
% clgroup -a group1 nodea nodeb nodec
To modify group "group1" to replace node "nodec" with node "noded", the following
command can be used:
% clgroup -r group1 nodec
% clgroup -a group1 noded
A new group "group2" can be formed from all of the nodes in "group1" and the additional
nodes "nodex" and "nodey" using the command:
% clgroup -a group2 group1 nodex nodey
The member nodes of group "group1" may be listed by using:
