Release NotesΒΆ

Scyld ClusterWare Release v12.2.0 is the latest update to Scyld ClusterWare.

For the most up-to-date product documentation and other helpful information, visit the Penguin Computing Support Portal. In particular, the most recent ClusterWare documentation can be found on the Penguin Computing Support Portal at https://www.penguinsolutions.com/computing/support/documentation/. The most recent version will accurately reflect the current state of the ClusterWare yum repository of RPMs that you are about to install.

Release Notes for Scyld ClusterWare:

  • Note that this will be the last release to support RHEL 7 and CentOS 7 head nodes. Images for those operating systems will continue to work and be supported in later releases, but starting with 12.3.0 ClusterWare will require an el8 or el9, including RHEL, Rocky, or CentOS Stream operating system on the head node(s).

  • This release introduces a new Role-Based Access Control (RBAC) system. The system allows for restricting the capabilities of individual administrator accounts by assigning them to specific roles. These restrictions are currently scoped across the entire cluster meaning an administrator could be restricted from modifying all images or rebooting all nodes. Further improvements will allow for object-based restrictions where an administrator may have complete control over certain images or nodes and not others.

    See _rbac_intro for additional details.

    Note that when upgrading to 12.2.0 or later any existing administrator accounts will be automatically marked as FullAdmins to mimic their current capabilities.

  • Partitioning using the ignition tool has been augmented to allow for installing the grub bootloader after partitioning and file system creation resulting in full deployments to local disks. The recommended best practice is still to use ephemeral compute node images that are reset on every boot, but local deployments allow cluster administators to use ClusterWare to manage long lived infrastructure nodes.

    Look for _bootloader and _ignition in _reserved_attributes for details.

  • This release also includes a new Mosquitto message broker service on every head node. The head node will publish events whenever primitives such as images or nodes are created or destroying. Cluster administrators can also create state maps and receive MQTT events whenever nodes move between states. The current MQTT topic layout is subject to future change.

  • This release includes preliminary Redfish support. ClusterWare can be configured to use a node's Redfish interface for power management. Additionally, ClusterWare now provides a standard Redfish interface itself, which aggregates the interfaces of all Redfish managed nodes.

Important

Upgrading from ClusterWare 11 to 12 may be more complicated than upgrading from ClusterWare 11 to a newer version of ClusterWare 11. Older clusters using Couchbase as the ClusterWare database must upgrade to the etcd database backend prior to the upgrade, and /etc/yum.repos.d/clusterware.repo may potentially require simple modifications to access the new Penguin Computing ClusterWare 12 repo. Please see Updating ClusterWare 11 to ClusterWare 12 for details.

Important

Updating from 12.0.1 and earlier to 12.1.0 requires reconfiguration of the Influx/Telegraf monitoring stack. The following command can be used to update the necessary config files: /opt/scyld/clusterware/bin/influx_grafana_setup --tele-env, followed by systemctl restart telegraf. All data will persist through the upgrade.

See Changelog & Known Issues for a full history of ClusterWare releases, and Known Issues And Workarounds for a summary of notable known current issues.