Thread: Kusu provisioning procedure

Package based provisioning

• call addhost and specify group, rack number:

*addhost listening /var/log/messages to discover the DHCP discover message. (addhost.py::scanSyslogForPXEReqs)
*Node was power on and broadcast a DHCP discover.
*addhost got the DHCP discover message in /var/log/messages;
*create entries in DB and then modify the lease file. (addhost.py::scanSyslogForPXEReqs)
*addhost call its plugins, the add methods to generate the tftp configs ...etc
*Node then could get IP address through DHCP request.
*Node then could download TFTP config files.
*Node downloading kernel, images, ks files, installing, reboot, run post scripts.
*When you quite addhost, it will call all its plugins, the finished methods to sync files...etc

• call customized anaconda which we hacked in initrd.

• customized anaconda call nodeinstaller to preparing kickstart for standard anaconda. packages/kusu-nodeinstaller/src/bin/nodeinstaller

*Download NII
*Set Timezone
*Format disk and mount filesystems
*Prepare SSH key
*Prepare root password
*Download post scripts

• call standard anaconda with the new generated kickstart file.

Image& diskless based provisioning

• addhost and pxe boot steps are same with above.

• The hacked init file: packages/initrd-templates/etc/imageinit.py

*Loading the required modules
*Bringing the NICs up to request an IP address
*Downloading the NII
*Configuring the NICs
*Download images and unpack it.
*Partitioning the disks (If needed)