Changed
- ConfigMap: Add lameduck of 5 seconds to health check (#191).
Connectivity
If you are upgrading from chart version lower than 3.0.0, it is neccessary to perform an intermediate upgrade to chart version 3.0.0 as this version contains a required migration process for enabling the use of seccompProfiles.
Changes
- Align with upstream chart version 2.16.4 (Changes in upstream repository)
- Change default
PodDisruptionBudgetto move frommaxUnavailable: 1tomaxUnavailable: 25%for better scaling
Removed
- Deployment: Drop static
replicas, managed by HPA. (#188)
- Deployment: Drop static
Changed
- Change
PodDisruptionBudgetto move frommaxUnavailable: 1tomaxUnavailable: 25%for better scaling
- Change
Changed
- Change PodDisruptionBudget to move from
maxUnavailable: 1tomaxUnavailable: 25%for better scaling
- Change PodDisruptionBudget to move from
Fixed
- Stop targeting default backend pods with the controller service. (#402)
This release upgrades kong to release 3.1.1 which contains breaking changes. Please consider reading the upstream Breaking changes documentation before upgrading. Users of the kong enterprise should read the Kong Gateway changelogs of releases 3.0.0.0 onwards up to 3.1.1.3.
Updates
- Align with upstream chart version 2.16.2 (Changes in upstream repository)
- Update kong/kong to 3.1.1. Please read Breaking changes notes
- Execute enterprise tests with kong-gateway container image version 3.1.1.3-debian
- Update kong ingress controller to 2.8.1
- Update custom resource definitions to those used by kong ingress controller [2.8.1]
Changes
- Breaking The
minReadySecondsvalue previous found atdeployment.kong.minReadySecondsis now configured at pathdeployment.minReadySeconds. - Increase default requests/limits to 1 cpu and 2G memory
- Remove startupProbe from kong ingress controller. More information
Added
- Align to upstream: Allow configuring the default backend.
Changes
- Add instance label to default topologySpreadConstraints label selector
- Update kong/kubernetes-ingress-controller to 2.8.1.
- Suggest using giantswarm/kong-gateway:2.8.2.3-debian as enterprise container image.
- Use
/readyzpath for ingress controller readiness check. - Update kubernetes api token secret with projected volume. This should fix problems with empty or expired ServiceAccount tokens.