A netboot=ipxe node is two DHCPv6 clients. The firmware solicits first and takes the Microsoft-signed shim over TFTP; the iPXE it chainloads solicits again with a DUID built from the same vpd.uuid in RFC 4122 order and takes the boot script of the node over HTTP. Its sibling case makedhcp_dhcpv6_boot_intent defines its node netboot=grub2, so no case read the pair, and a cell could report green with every iPXE line of dhcp.pm broken. makedhcp_dhcpv6_ipxe_boot_intent reads the rendered kea-dhcp6.conf. It counts the reservations keyed on the two expected DUIDs, compares both entries byte for byte, asserts the two DUIDs differ, and moves the address between the clients with mac.duid to say WHICH client gets WHICH file. A grub2 arm and a return to iPXE are the discriminator: a case that cannot tell the two methods apart only measures that a reservation exists. The loader check is first. dhcp6_bootfile_url_for_node returns empty when the UEFI loader is not under the tftpdir of the node and makedhcp still exits 0, so every later assertion would pass vacuously on a management node with no ipxe-xcat. Measured against the rendered reservations of this tree: reverting dhcp6_chainloads to return 0 takes the count from 2 to 1 and leaves the first reservation and the grub2 arm passing; deleting the "if ($client eq 'ipxe')" arm of dhcp6_bootfile_url_for_node turns the second URL into the TFTP shim and leaves nine of the eleven value assertions passing. ipxe-xcat is packaged for x86_64 only, so a bundle must name this case for an x86_64 cell. Signed-off-by: Daniel Hilst <392820+dhilst@users.noreply.github.com>
xCAT
xCAT is a toolkit for deployment and administration of clusters of all sizes.
The xCAT sunset was only a quick eclipse
Dear xCAT Community,
The xCAT sunset has changed course. VersatusHPC has been invited to join the xCAT Consortium, and future development will move toward direct upstream contributions coordinated with the Consortium and its existing member companies.
That matters most for Enterprise Linux 10 (EL10). EL10 support is coming to xCAT, restoring a future operating-system path for sites that still rely on xCAT. This is an important change from the previous sunset guidance, where the lack of an EL10 path was one of the strongest reasons to move away from xCAT.
The xCAT Consortium continues to recommend Confluent as the long-term successor to xCAT, and that remains the Consortium position. Users planning new cluster-management deployments should evaluate Confluent and its xCAT comparison documentation.
At the same time, xCAT is no longer sunsetted. The Consortium and participating companies will continue updating xCAT while there is community and user demand for it.
In summary:
- xCAT development is continuing upstream through the Consortium and participating companies.
- Enterprise Linux 10 support is coming.
- Confluent remains the Consortium-recommended successor and migration path.
- xCAT updates will continue while there is community and user demand.
We want to thank the xCAT Consortium and community for keeping this project moving. The sun went behind the moon for a moment, but xCAT is still here.
For more information on Confluent and how to get started, please visit the Confluent: Project Page, Documentation or Confluent vs xCAT comparison.
With thanks,
The xCAT Consortium
Documentation
xCAT Documentation is hosted on Read The Docs: https://xcat-docs.readthedocs.io
Status
| xCAT Version | Build Status |
|---|---|
| Latest (master branch) | |
| Stable (latest release) |
Looking for older versions?
Open Source License
xCAT is made available under the EPL license: https://opensource.org/licenses/eclipse-1.0.php
Developers
Want to help? Check out the developers guide!