Releases: kubernetes/ingress-nginx
Nginx: 0.9.0-beta.2
Image: gcr.io/google_containers/nginx-ingress-controller:0.9.0-beta.2
New Features:
- New configuration flag
proxy-set-headersto allow set custom headers before send traffic to backends. Example here - Disable directive access_log globally using
disable-access-log: "true"in the configuration ConfigMap. - Sticky session per Ingress rule using the annotation
ingress.kubernetes.io/affinity. Example here
Complete changelog here
GCE: 0.9.1
Image: gcr.io/google_containers/glbc:0.9.1
Changelog:
- Default balancing mode UTILIZATION -> RATE: We now try to create backends with balancing mode RATE first, and if that fails, create them with balancing mode UTILIZATION.
Getting Started:
Kick the tires by creating an ingress. You can also perform a dry run on any machine where kubectl works against a Kubernetes cluster by invoke the release binary with glbc --running-in-cluster=false. A dry run will not create any cloud resources.
Make sure you satisfy the pre-requisites, consult these docs for troubleshooting and see the wishlist if you'd like to contribute to the next release.
GCE: 0.9.0
Image: gcr.io/google_containers/glbc:0.9.0
Changelog:
- Health Checks correctly match named ports: We were previously ignoring the port name on a container when looking for readiness probes to convert into GCE health checks.
Getting Started:
Kick the tires by creating an ingress. You can also perform a dry run on any machine where kubectl works against a Kubernetes cluster by invoke the release binary with glbc --running-in-cluster=false. A dry run will not create any cloud resources.
Make sure you satisfy the pre-requisites, consult these docs for troubleshooting and see the wishlist if you'd like to contribute to the next release.
Nginx: 0.9.0-beta.1
Image: gcr.io/google_containers/nginx-ingress-controller:0.9.0-beta.1
New Features:
- SSL Passthrough
New Flag --publish-service that set the Service fronting the ingress controllers - Ingress status shows the correct IP/hostname address without duplicates
- Custom body sizes per Ingress
- Prometheus metrics
Breaking changes:
- Flag --nginx-configmap was replaced with --configmap
- Configmap field body-size was replaced with proxy-body-size
Complete change is here
GCE: 0.9.0-beta.1
Image: gcr.io/google_containers/glbc:0.9.0-beta.1
Changelog:
- Smarter URL map updates: Compare the retrieved URL Map and the current in-memory URL Map, if there's no change, don't send an update.
- Smarter refcounting of default backend: If there's an Ingress pointing at a Service without a nodeport, requests sent to the Ingress IP should hit the default backend. For this to work, we can't delete the default backend till there are no more loadbalancers.
Getting Started:
Kick the tires by creating an ingress. You can also perform a dry run on any machine where kubectl works against a Kubernetes cluster by invoke the release binary with glbc --running-in-cluster=false. A dry run will not create any cloud resources.
Make sure you satisfy the pre-requisites, consult these docs for troubleshooting and see the wishlist if you'd like to contribute to the next release.
GCE: 0.9.0-alpha.1
Image: gcr.io/google_containers/glbc:0.9.0-alpha.1
Changelog:
- Better logging and events: Clarity around health checks, node ports etc.
- Default backend is created lazily: Previously, the pattern of creating a default backend was creating delay in cleanup. Refactored so we don't spew bad logs trying to recreate it when there are no ingresses.
- Decreased health check interval: LB health checks were created with a 1/s interval. It will now be a minimum of 1/minute.
Getting Started:
Kick the tires by creating an ingress. You can also perform a dry run on any machine where kubectl works against a Kubernetes cluster by invoke the release binary with glbc --running-in-cluster=false. A dry run will not create any cloud resources.
Make sure you satisfy the pre-requisites, consult these docs for troubleshooting and see the wishlist if you'd like to contribute to the next release.