-
Bug
-
Resolution: Won't Do
-
Undefined
-
None
-
None
-
2
-
False
-
None
-
False
-
-
-
Tracing Sprint # 259, Tracing Sprint # 260, Tracing Sprint # 261
I'm working on something using the OpenTelemetry Collector API provided by this repository and I found some configuration fields were not marshaled properly.
So, for an OpenTelemetry Collector like this:
apiVersion: opentelemetry.io/v1beta1
kind: OpenTelemetryCollector
metadata:
name: instance
namespace: open-cluster-management-observability
spec:
config:
exporters:
debug:
processors:
receivers:
jaeger:
protocols:
grpc:
otlp:
protocols:
grpc:
http:
service:
pipelines:
traces:
exporters:
- debug
processors: []
receivers:
- otlp
- jaeger
mode: deployment
I can see the protocols field from the otlp receiver is dropped.
If we try directly in a script, it doesn't reproduce. But it can be reproduced by creating the resource first in Kubernetes, retrieving it, and doing the marshaling.
This seems to be expected from the changes introduced in https://github.com/open-telemetry/opentelemetry-operator/pull/2444
Upstream issue: https://github.com/open-telemetry/opentelemetry-operator/issues/3281