v0.11

There should be no breaking change. 🎉

• Fixed an error about a missing index on the API server caused by the affinity plugin, when neither the Kubernikus or ESX integration were enabled.

v0.10

There should be no breaking change. 🎉

• Adjusted leader election. Config Maps are no longer used for leader election and non-leaders will now try to become leader every 5s instead of every 2s.

v0.9

Breaking change 💥:

• The maintenance-controller is now less strict about profiles in cloud.sap/maintenance-profile, which are not present in the configuration. These are now silently ignored (previously such nodes where skipped entierly).

Other changes 🎉:

• Certain dates can now be excluded from the timewindow check plugin

v0.8

There should be no breaking change. 🎉

• Fix deleted nodes being endlessly requeued and thereby cluttering the log with stacktraces

v0.7

There should be no breaking change. 🎉

• The affinity plugin now only considers nodes, which are in-maintenance caused by the same profile. This resolves some unnecessary blocking.
• Fixed a nil pointer deref in the affinity plugin.
• Added kubernikus integration, which checks for kubelet/API server version mismatches and can delete nodes from OpenStack.
• Added nodecount plugin, which checks that at least a specified amount of nodes is present in a cluster.

v0.6

There should be no breaking change. 🎉

• The stagger check plugin now supports a parallel config option, which enables nodes to "race" for multiple leases. The configured lease name is appended with -INDEX
• The leases used by the stagger check plugin are now only grabbed by nodes, if they actual do a state transition. This resolves some unnecessary blocking that could occur.
• Added the affinity check plugin, which helps with scheduling pods on operational nodes to minimize downtime.

v0.5

Breaking change 💥:

• The slackThread notification plugin now requires a title to be configured, which is displayed directly in the slack channel. The messages generated by each node appear as replies.

Other changes 🎉:

• Renamed the leader election lease, the old one can be deleted
• The ESX controller now reloads parts of its configuration on each reconciliation, so pod restarts may be avoided
• Updated to go 1.17

v0.4

Breaking change 💥:

• The separator for specifying multiple profiles has changed from - to --

Other changes 🎉:

• Events generated by the maintenance-controller now appear in kubectl describe node
• Added slack thread notification plugin

v0.3

There should be no breaking change. 🎉

• Added stagger check plugin
• Added an optional sub-controller, which deals with the maintenance of ESX hosts (to be enabled with: --enable-esx-maintenance)

v0.2

There should be no breaking change. 🎉

• Added node condition check plugin
• Multiple maintenance profiles can now be specified for node using the cloud.sap/maintenance-profile=PROFILE1-PROFILE2-... label