"Failed to get ports for pod" logged at ERROR level for most pods

[benbc]

I'm using datadog-agent 6.1.0 on Kubernetes v1.8.9, running on GKE.

When the agent starts up I see several logs like this:

[ AGENT ] 2018-03-26 09:31:40 UTC | ERROR | (kubelet.go:170 in createService) | Failed to get ports for pod fluentd-gcp-v2.0.9-976xf

Looking at the code (

datadog-agent/pkg/collector/listeners/kubelet.go

Line 170 in 614a3ff

log.Errorf("Failed to get ports for pod %s", podName)

) I see that this error log is made for every container that doesn't have any ports specified.

I don't think that this is an error condition, or even something that should be a warning.

[azzeddinefaik]

I got the same issue here ? any way to solve this ?

[azzeddinefaik]

[ AGENT ] 2018-04-23 12:50:40 UTC | ERROR | (kubelet.go:170 in createService) | Failed to get ports for pod fluentd-6bzxm

[jimmyraywv]

Is this a bug or at least an error with a known fix? I am trying to understand the source if this error and how to remedy it. I am using datadog:6.1.4.

(kubelet.go:170 in createService) | Failed to get ports for pod calico-node-... (kubelet.go:170 in createService) | Failed to get ports for pod calico-node-... (kubelet.go:170 in createService) | Failed to get ports for pod fluent-bit-... (kubelet.go:170 in createService) | Failed to get ports for pod prometheus-alertmanager-...

[CharlyF]

Hey all,

This was indeed set as Error while it should just be a Debug log.
It happens if for instance we don't find ports in pod spec of the pod.
The PR that fixed this is #1490 and the fix is embedded in the version 6.2 of the agent that we released a few days ago (will be available at datadog/agent:latest).

Could you give this a try and let us know if that helps ?
Best Regards,