Using confluent discovery to feed xCAT configuration
While xCAT has its own discovery mechanism, in some scenarios confluent discovery may be desirable in an otherwise xCAT environment. For example, confluent does not require a dynamic range and does not require net-snmp-perl. Additionally, confluent supports standby power discovery in a number of scenarios, enabling management and console access prior to turning on the system.
Preparing for PXE mac collection
In order to direct confluent to retain PXE mac addresses, simply set the net.bootable
attribute to 1 and ensure that discovery.policy
includes PXE:
# nodegroupattrib everything discovery.policy=permissive,pxe net.bootable=1
Perform normal confluent discovery
Confluent discovery may be done against the BMC device or against PXE attempts. If using the network switch as a reference for discovery, be sure to use the dedicated management port switch and port for standby power discovery, and the network boot port switch and port if doing PXE. Detailed procedures on a few strategies for discovery are:
- Discovering systems attached to ethernet switches
- Discovering systems in enclosures
- Discovering systems manually
Populating xCAT with node information
Confluent has a command called confluent2xcat
to export confluent data to xCAT. It can generate one of two files.
In order to generate a ‘stanza’ format file for mkdef (pulling in confluent group membership and other information):
# confluent2xcat d1-d8 -o nodes.stanza
In order to generate a CSV of node and mac addresses suitable for tabrestore mac:
# confluent2xcat d1-d8 -m mac.csv