0

I am trying to deploy Influx DB on Kubernetes. Using the official Influx DB Helm chart.

I am deploying this Helm chart with the following values file:

persistence:
    enabled: true
    size: 5Gi

ingress:
        enabled: true
        # tls: true
        hostname: influxdb.slackdog.space
        annotations: 
            kubernetes.io/ingress.class: "gce"
            external-dns.alpha.kubernetes.io/hostname: influxdb.slackdog.space.

service:
       type: NodePort

When I deploy the Helm chart one of the Backend services stays in an UNHEALTHY state. I have left the deployment sitting for days with no status change.

Here is the ingress controller in question

Name:             slack-dog-influxdb-influxdb
Namespace:        slack-dog-influxdb
Address:          xx.xxx.xxx.xx # Note: external IP
Default backend:  default-http-backend:80 (xx.xx.x.xx:8080) # Note internal cluster IP
Rules:
  Host                     Path  Backends
  ----                     ----  --------
  influxdb.slackdog.space  
                           /   slack-dog-influxdb-influxdb:8086 (<none>)
Annotations:
  backends:         {"k8s-be-30983--01721a5ee78653ec":"HEALTHY","k8s-be-32128--01721a5ee78653ec":"UNHEALTHY"}
  forwarding-rule:  k8s-fw-slack-dog-influxdb-slack-dog-influxdb-influxdb--01721a50
  target-proxy:     k8s-tp-slack-dog-influxdb-slack-dog-influxdb-influxdb--01721a50
  url-map:          k8s-um-slack-dog-influxdb-slack-dog-influxdb-influxdb--01721a50
Events:
  Type    Reason  Age   From                     Message
  ----    ------  ----  ----                     -------
  Normal  ADD     35m   loadbalancer-controller  slack-dog-influxdb/slack-dog-influxdb-influxdb
  Normal  CREATE  35m   loadbalancer-controller  ip: xx.xxx.xxx.xx # Note: external IP

I have run a bash terminal in the deployed InfluxDB pod and successfully pinged the localhost:8086/ping endpoint, which is used for health checks in the deployment.

I tried adding a firewall rule to allow traffic in on port 8086 but this seams to have no effect:

enter image description here

What could be causing this behavior? Is there any manual tweaking I need to perform in the GCP to get things working?

Noah Huppert
  • 101
  • 3
  • Can you verify [these list of IP ranges](https://cloud.google.com/load-balancing/docs/health-checks#network_load_balancing) mentioned in the help center article allowed in the firewall rules?. Also I have seen a [similar issue](https://github.com/jetstack/kube-lego/issues/27), where correcting the path resolved the issue. See if that helps you. – Digil Aug 17 '18 at 20:25
  • @Digil Thanks for the response. I ended up realizing that the Influx DB service only needs to be accessed from within the cluster, so I removed the Ingress resource which this problem was related to. We will never know what the problem actually was. It would be nice if GKE provided a bit more information about what was going wrong when something like this occurs. It is hard to debug all the different GCP resources which GKE stands up to complete its tasks. – Noah Huppert Aug 20 '18 at 03:17
  • The ingress resource provides traffic logs through stackdriver, which is a good place to start when you are experiencing issues there. As for the rest, you need to perform most of your [debugging from within the cluster directly](https://kubernetes.io/docs/tasks/debug-application-cluster/debug-service/) – Patrick W Aug 22 '18 at 21:53

0 Answers0