Introduction
There are situations where configuring a dedicated cloning network on your headnode HA is required, or would be useful. Some examples of such situations are:
- The cluster doesn’t have an Ethernet-based management network, and only uses the InfiniBand fabric.
- Pushing the headnode cloning traffic for the HA setup process through the internal network may not be desirable due to bandwidth concerns.
The dedicated cloning network may be a directly-connected cable running between the two headnodes.
The steps that follow would generally be used after the initial run of the cmha-setup script and before booting the secondary headnode into rescue mode for cloning.
Configuring a dedicated cloning network
To start with, create a clone of your existing internal net. Lacking any imagination, we will set it to clonenet:
# cmsh % network % clone internalnet clonenet % set nodebooting no % set managementallowed no
Remember to modify the base address, broadcast address, dynamic range start, dynamic range end, and domain name fields to something else that is appropriate. These cannot be the same as the existing internalnet.
Once completed, commit the changes.% commit
On the primary active master, we need to configure some interfaces for the clonenet.
In the example configuration below, ethX should be replaced with a real network interface (for example, eth3) on the system.
On the primary active master headnode:% device use master % interfaces % add physical ethX % set ip address X.X.X.X % set network clonenet % commit % .. % add alias ethX:cmha0 % set startif active % set ip X.X.X.X % set network clonenet % commit % device use master2 <-----(The secondary headnode) % interfaces % add physical ethX % set ip address X.X.X.X % set network clonenet % commit % .. % add alias ethX:cmha0 % set startif active % set ip X.X.X.X (Same IP as master ethX:cmha0) % set network clonenet % commit % network use clonenet % set nodebooting yes % set managementallowed yes % commit
You may need to reboot the active primary headnode after these networking changes.
Now reboot the secondary headnode and set it to PXEboot off the appropriate Ethernet interface in clonenet, and you should be able to boot the system into rescue mode.
Now run the cm-clone-install and select the appropriate Ethernet interface.