Changelog

⚠️ Breaking changes ⚠️ `outputRefs` is now deprecated!

For clarity we introduce:

use globalOutputRefs for ClusterOutputs
use localOutputRefs for Outputs

Why?

There were a lot of confusion about the difference and precedence between Outputs and ClusterOutputs so we reflect the type in the Flow and ClusterFlow parameters.
Moreover, ClusterFlow and ClusterOutput will become cluster scoped resources in later releases.

What does it mean?

You need to change your custom resources according to the new parameters.

What happens if I already upgraded but I don't want to change the parameters.

You can downgrade to Logging Operator 3.5.1 and everything should work the same as before.

Example: There is a Flow with 1 Output (elasticsearch) and 1 ClusterOutput (s3)

Old syntax

kind: Flow
metadata:
  name: s3-flow
spec:
  filters:
    - tag_normaliser: {}
  match:
    - select:
        labels:
          app.kubernetes.io/name: log-generator
  outputRefs:
    - elasticsearch
    - s3

New syntax

kind: Flow
metadata:
  name: s3-flow
spec:
  filters:
    - tag_normaliser: {}
  match:
    - select:
        labels:
          app.kubernetes.io/name: log-generator
  globalOutputRefs:
    - s3-output
  localOutputRefs:
    - elasticsearch

Http Output json_array parameter support #566
Splunk Documentatio #563
Sumologic documentation #561
AWS Process Credentials Support #558
Suppress_type_name support in elasticsearch output #556
Separate cluster output refs from output refs #550
Redis Output Plugin Support #549

Fluentd

New FluentD version 1.11.1 -> 1.11.2

Updated Plugins

fluent-plugin-elasticsearch 4.1.1 -> 4.1.2
fluent-plugin-kafka 0.13.1 -> 0.14.2

New Plugins

fluent-plugin-redis 2.4.0

Fluentbit

New Fluentbit version 1.5.2 -> 1.5.4
Milestone 3.6.0
Detailed changes: 3.5.0...3.6.0
Thanks for all the contributors: @ahma @asdwsda @bshifter @pepov @eddycharly @tarokkk @siliconbrain

kube-logging/logging-operator 3.6.0 on GitHub