Redhat-opentelemetry-collector

[helio-frota]

I’ve learned a few new things:

• We can use the official OpenTelemetry helm chart and reference a different image (a custom build) and tag
• The redhat-opentelemetry-collector is UBI based, while the community one uses Alpine
• It’s possible to override the custom image and tag via helm arguments without modifying values.yaml (at least in the example I’m using)
  • FooBar image and tag:

    helm upgrade --install --dependency-update -n $NAMESPACE infra charts/infra 
    --values charts/infra/values.yaml --set-string opentelemetry-collector.image.repository=aaa 
    --set-string opentelemetry-collector.image.tag=123
    

    kubectl logs infra-otelcol-6894f7bcbd-88vj2
    Error from server (BadRequest): container "opentelemetry-collector" in pod "infra-otelcol-6894f7bcbd-88vj2"
    is waiting to start: image can't be pulled
    

  • Existing image and tag

    helm upgrade --install --dependency-update -n $NAMESPACE infra charts/infra 
    --values charts/infra/values.yaml --set-string 
    opentelemetry-collector.image.repository=ghcr.io/os-observability/redhat-opentelemetry-collector/redhat-opentelemetry-collector 
    --set-string opentelemetry-collector.image.tag=v0.113.0
    

    kubectl logs infra-otelcol-56fc4d8465-sp7zv
    2025-05-12T21:15:06.816Z	info	[email protected]/service.go:166	Setting up own telemetry...
    2025-05-12T21:15:06.816Z	warn	[email protected]/service.go:221	service::telemetry::metrics::address is being deprecated in favor of service::telemetry::metrics::readers
    2025-05-12T21:15:06.816Z	info	telemetry/metrics.go:70	Serving metrics	{"address": "10.244.0.28:8888", "metrics level": "Normal"}
    2025-05-12T21:15:06.816Z	info	builders/builders.go:26	Development component. May change in the future.	{"kind": "exporter", "data_type": "logs", "name": "debug"}
    2025-05-12T21:15:06.816Z	info	[email protected]/memorylimiter.go:151	Using percentage memory limiter	{"kind": "processor", "name": "memory_limiter", "pipeline": "logs", "total_memory_mib": 20000, "limit_percentage": 80, "spike_limit_percentage": 25}
    2025-05-12T21:15:06.816Z	info	[email protected]/memorylimiter.go:75	Memory limiter configured	{"kind": "processor", "name": "memory_limiter", "pipeline": "logs", "limit_mib": 16000, "spike_limit_mib": 5000, "check_interval": 5}
    2025-05-12T21:15:06.816Z	info	builders/builders.go:26	Development component. May change in the future.	{"kind": "exporter", "data_type": "traces", "name": "debug"}
    2025-05-12T21:15:06.817Z	info	builders/builders.go:26	Development component. May change in the future.	{"kind": "exporter", "data_type": "metrics", "name": "debug"}
    2025-05-12T21:15:06.817Z	info	[email protected]/service.go:238	Starting otelcol...	{"Version": "0.113.0", "NumCPU": 12}
    2025-05-12T21:15:06.817Z	info	extensions/extensions.go:39	Starting extensions...
    

Unless there’s something different in trustify-helm-charts compared to this example, we could hard-code redhat-opentelemetry-collector and document the additional arguments for the community version in the README.

[helio-frota]

Closing, I shared it here so it doesn't get lost in the chat