0.4.3

What's Changed

• discard not yet reconciliated node disruption by @pgoron in #73

New Contributors

• @pgoron made their first contribution in #73

Full Changelog: 0.4.2...0.4.3

0.4.2: Improve logging and reasons (#71)

Several changes:
- deadline exceeded: Feedback was that it was too general (looked like http error)
- disruption type message is clearer (display allowed types as well)
- log all rejections: before we logged only the "budget" rejections, now all rejections are logged.

0.4.1: Improve PVC phase handling (#70)

Now:
- ignore PVC that are pending
- raise a specific error if the phase is not expected

Since Status is a subresource, it has to be updated directly in the tests

0.4.0

What's Changed

• Improve deadline exceed message by @geobeau in #65
• Pin go correctly to 1.21 by @geobeau in #69
• Implement spec.type to Node Disruption API [#64] by @FatmaBouzghaia in #67

New Contributors

• @FatmaBouzghaia made their first contribution in #67

Full Changelog: 0.3.2...0.4.0

0.3.2

What's Changed

• Fix wrong value in labels by @geobeau in #62
• Properly set labels for disruption budgets by @geobeau in #63

Full Changelog: 0.3.1...0.3.2

0.3.1

What's Changed

• Fix bugs on metrics by @geobeau in #61
• Fix empty node name in watched nodes by @geobeau in #60

Full Changelog: 0.3.0...0.3.1

0.3.0

What's Changed

• Add support for prometheus metrics in helm chart by @geobeau in #53
• Several metrics improvements by @geobeau in #54
• Add full API documentation by @geobeau in #55
• Add metrics for budget by @geobeau in #56
• Update CRD with new version of kubebuilder by @geobeau in #57
• Remove healthURL to use healthHook only by @geobeau in #58
• Prepare bump to 0.3.0 by @geobeau in #59

Full Changelog: 0.2.3...0.3.0

0.2.3: Add metrics to node disruption (#52)

Add the following metrics:
- state: reflect the state of the node disruption
- created and deadline: report as value the creation timestamp and the deadline
- impact nodes: report for each nodes a metric, helpful for interesection

Unittesting of metrics in go is not easy, it will be part of a future change

0.2.2: Add a proper content type to Healthhook (#51)

Some application require the Content-Type be set to json
to parse json. So now we provide it by default

0.2.1: Fix flaky test when reconciler start before previous one still up

between test, the reconciler is stopped and started again.
Stop is not instant but we started a new instance right away.
Since it bind a port, the new one crashed because the old
one didn't release it yet. Instead we wait for the manager to
be properly down before starting a new one