After setting up a Kubernetes service running an FTP application with port mapping 21 inside AKS, I noticed that port 21 is being probed by some kind of process. The following is being logged from within my FTP application:
[I 2019-12-06 23:38:56] 10.244.0.1:56287-[] FTP session closed (disconnect).
[I 2019-12-06 23:38:56] 10.244.0.1:56330-[] FTP session opened (connect)
[I 2019-12-06 23:39:02] 10.244.0.1:56330-[] FTP session closed (disconnect).
[I 2019-12-06 23:39:02] 10.244.0.1:56376-[] FTP session opened (connect)
The VM this pod is on has IP range 10.244.0.0/24
There is no Container Insights or anything else configured
I don't believe this is Readiness or Liveness probes, as I have overridden those - but it does look a lot like some kind of generic TCP probe on whatever port is being configured in the port mapping?
The full dump of all pods gives
NAMESPACE NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
kube-system coredns-866fc6b6c8-5cgll 1/1 Running 0 86m 10.244.0.7 aks-agentpool-40501316-0 <none> <none>
kube-system coredns-866fc6b6c8-nrbk2 1/1 Running 0 93m 10.244.0.4 aks-agentpool-40501316-0 <none> <none>
kube-system coredns-autoscaler-5d5695b54f-82jkb 1/1 Running 0 92m 10.244.0.2 aks-agentpool-40501316-0 <none> <none>
kube-system kube-proxy-sk59f 1/1 Running 0 81m 10.240.0.4 aks-agentpool-40501316-0 <none> <none>
kube-system kubernetes-dashboard-6f697bd9f5-s7kgz 1/1 Running 0 92m 10.244.0.5 aks-agentpool-40501316-0 <none> <none>
kube-system metrics-server-566bd9b4f7-gqnw9 1/1 Running 1 93m 10.244.0.3 aks-agentpool-40501316-0 <none> <none>
kube-system tunnelfront-855c64f898-xhk6m 1/1 Running 0 92m 10.244.0.6 aks-agentpool-40501316-0 <none> <none>
test test-b8487b946-z6tlb 1/1 Running 0 9m24s 10.244.0.13 aks-agentpool-40501316-0 <none> <none>
I must be missing something obvious here. What is constantly polling my application every 6 seconds? On a node with IP range 10.244.0.0/24, what is the thing that is being created at 10.244.0.1?
If you are having a service of type LoadBalancer
mapped to this pod - this would be the Azure Load Balancer probes doing the health check to determine if the service is alive.