Application-gateway-kubernetes-ingress: Application Gateway stops working when using a long domain name

Created on 27 Mar 2019 · 5Comments · Source: Azure/application-gateway-kubernetes-ingress

Describe the bug
The application gateway stops working when a long domain name is used for an application. We have encountered the issue when using a domain name of 51 characters. Another domainname of 43 characters of the same deployment works fine. The exact minimum number of characters that breaks the Application Gateway has not been determined.

When the Application Gateway breaks, a psping on port 80 shows that the Application Gateway does not respond anymore. Browsers sessions report ERR_CONNECTION_TIMED_OUT. After a couple of minutes the Health Status in _Diagnose and solve problems_ changes to Unavailable. Removing the rules results in the Application Gateway working again. The configuration of the Application Gateway reflects the domain name in the rules created.

To Reproduce
Deploy an application with a domain name of 51 characters.

Ingress Controller details

Output of kubectl describe <ingress controller>
Name: ******-*****-ingress-azure-798c6d5676-wtw7b Namespace: default Priority: 0 PriorityClassName: <none> Node: aks-agentpool-35064155-0/10.0.0.4 Start Time: Wed, 27 Mar 2019 10:39:02 +0100 Labels: aadpodidbinding=******-******-ingress-azure app=ingress-azure pod-template-hash=798c6d5676 release=******-****** Annotations: <none> Status: Running IP: 10.0.0.32 Controlled By: ReplicaSet/******-******-ingress-azure-798c6d5676 Containers: ingress-azure: Container ID: docker://8e374c0410611c52cc7482ced218fdbdc8ab9e33aa131e7ffee68085c8f459ae Image: mcr.microsoft.com/azure-application-gateway/kubernetes-ingress:0.2.0 Image ID: docker-pullable://mcr.microsoft.com/azure-application-gateway/kubernetes-ingress@sha256:c8e7b9182e9c417bc6ac0470a3276f80314e78b0f752ffd70216f390fbd15f2d Port: <none> Host Port: <none> State: Running Started: Wed, 27 Mar 2019 10:39:05 +0100 Ready: True Restart Count: 0 Environment Variables from: ******-******-cm-ingress-azure ConfigMap Optional: false Environment: KUBERNETES_PORT_443_TCP_ADDR: aks-********.hcp.westeurope.azmk8s.io KUBERNETES_PORT: tcp://aks-********.hcp.westeurope.azmk8s.io:443 KUBERNETES_PORT_443_TCP: tcp://aks-********.hcp.westeurope.azmk8s.io:443 KUBERNETES_SERVICE_HOST: aks-********.hcp.westeurope.azmk8s.io Mounts: /var/run/secrets/kubernetes.io/serviceaccount from ******-******-sa-ingress-azure-token-c6bwp (ro) Conditions: Type Status Initialized True Ready True ContainersReady True PodScheduled True Volumes: ******-******-sa-ingress-azure-token-c6bwp: Type: Secret (a volume populated by a Secret) SecretName: ******-******-sa-ingress-azure-token-c6bwp Optional: false QoS Class: BestEffort Node-Selectors: <none> Tolerations: node.kubernetes.io/not-ready:NoExecute for 300s node.kubernetes.io/unreachable:NoExecute for 300s Events: <none>
Azure support ticket: [REG:119032125000498] The health status of the AppGW changes to Unhealthy in a few minutes after restart.

Source

MarkDNL

Most helpful comment

This is a fix in AppGW v2. It is already committed and should be released next month.

akshaysngupta on 27 Mar 2019

👍2

All 5 comments

After doing some more research, this is not an issue with application-gateway-kubernetes-ingress. I entered the same configuration via the Azure Portal, which also resulted in the Application Gateway becoming unavailable. This should be checked/validated by the Application Gateway,

MarkDNL on 27 Mar 2019

Yes, this is a bug in the Application Gateway (v2?) We have also encountered the bug,
which is related to the length of the rule name. We are already working with Microsoft Support to solve this.