Exclude annotation is not processed when pod has multiple containers

[Scukerman]

Bug Report

Describe the bug
fluentbit.io/exclude annotation doesn't work on pod if the pod has multiple containers.
fluentbit.io/exclude_stdout-nginx also doesn't work.

To Reproduce

• Enable K8S-Logging.Exclude in the fluentbit config.
• Annotate a pod with at least two containers with the command (kubectl annotate po -l app=demo-server fluentbit.io/exclude='true', adjust it for your environment).
• Delete the pod so fluent bit could reload the configuration and apply the annotations.

Expected behavior
All pod logs of all containers should be excluded from processing

Screenshots

Your Environment

• Version used: 1.4.3
• Configuration: https://pastebin.com/muDaDxB4
• Environment name and version (e.g. Kubernetes? What version?): kubernetes 1.17.7
• Server type and version: bare metal
• Operating System and version: ubuntu
• Filters and plugins: kubernetes

[github-actions]

This issue is stale because it has been open 30 days with no activity. Remove stale label or comment or this will be closed in 5 days.

[github-actions]

This issue was closed because it has been stalled for 5 days with no activity.

[connorworkman]

Bug still exists as of fluent-bit v1.9.4

[Scukerman]

I wish they could fix it in 3.0.0 but I won't check it because I switched to another solution

[brownz11]

I know this is an old issue, but it's one of the first Google results.

In my case (v2.1.0) it's not a bug with the exclude, but the buffer setting for the Kubernetes Filter was too low. So it would get a response, but discard it and have no annotations to do the filtering on. Luckily it's an easy fix, just increase the Buffer_Size...

    [FILTER]
        Name                kubernetes
        Match               kube.*
        Kube_URL            https://kubernetes.default.svc.cluster.local:443
        Buffer_Size         1M #Default is 32k which is was not enough for my cluster where we had sidecars
        K8S-Logging.Exclude On