strategy: production = does not have a named port

[Gary Golub]

I've deployed jaeger using the operator into an AKS cluster.  I configured it using the "simple" strategy which works fine.  I then tried changing it to the "production" strategy and I get errors like this:

error: Pod 'jaeger-prod-collector-996cc4f9-ghgxp' does not have a named port 'pod/jaeger-prod-query-76bc8b6f74-wbwpw'

Here is my configuration:

apiVersion: jaegertracing.io/v1

kind: Jaeger

metadata:

  name: jaeger-prod

  namespace: observability

spec:

  strategy: production

  storage:

    type: elasticsearch

    options:

      es:

        server-urls: http://elasticsearch-master.elastic-system.svc.cluster.local:9200

  collector:

    options:

      log-level: debug

    maxReplicas: 5

    resources:

      limits:

        cpu: 100m

        memory: 128Mi

I'm not clear what the message means nor what I may be missing in my configuration.  I can't find any errors in the logs nor any examples showing a working "production" configuration.

Can anyone tell me what I'm missing and, perhaps, provide a concrete example of a configuration using this strategy?

Thanks in advance.

[Juraci Paixão Kröhling]

Can you reproduce it using minikube? If so, please open an issue in the
following repository: https://github.com/jaegertracing/jaeger-operator

- Juca.

On 16.03.21 22:19, Gary Golub wrote:
> I've deployed jaeger using the operator into an AKS cluster.  I
> configured it using the "simple" strategy which works fine.  I then
> tried changing it to the "production" strategy and I get errors like this:
>
> error: Pod 'jaeger-prod-collector-996cc4f9-ghgxp' does not have a named
> port 'pod/jaeger-prod-query-76bc8b6f74-wbwpw'
>
> Here is my configuration:
>
> apiVersion: jaegertracing.io/v1
> kind: Jaeger
> metadata:
>   name: jaeger-prod
>   namespace: observability
> spec:

> *strategy: production*

>   storage:
>     type: elasticsearch
>     options:
>       es:
>         server-urls:
> http://elasticsearch-master.elastic-system.svc.cluster.local:9200
>   collector:
>     options:
>       log-level: debug
>     maxReplicas: 5
>     resources:
>       limits:
>         cpu: 100m
>         memory: 128Mi
>
> I'm not clear what the message means nor what I may be missing in my
> configuration.  I can't find any errors in the logs nor any examples
> showing a working "production" configuration.
>
> Can anyone tell me what I'm missing and, perhaps, provide a concrete
> example of a configuration using this strategy?
>
> Thanks in advance.
>

> --
> You received this message because you are subscribed to the Google
> Groups "Jaeger Tracing" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to jaeger-tracin...@googlegroups.com
> <mailto:jaeger-tracin...@googlegroups.com>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/jaeger-tracing/d67fe939-705c-4712-9875-b2dceac45531n%40googlegroups.com
> <https://groups.google.com/d/msgid/jaeger-tracing/d67fe939-705c-4712-9875-b2dceac45531n%40googlegroups.com?utm_medium=email&utm_source=footer>.

[Gary Golub]

Sorry for the red herring.

I was able to reproduce it in OpenStack and Minikube but, in doing so, determined the problem was something dumb on my part.

I have this alias defined:

alias jaegerui='kubectl port-forward $(kubectl get pods -l app="jaeger" -o name -n $JAEGER_NAMESPACE | grep query) 16686:16686 -n $JAEGER_NAMESPACE'

When using the "production" strategy you end up with multiple pods rather than one as with the "simple" strategy.  That caused the alias to get back multiple values from "get pods" which didn't play well with the port-forward comment.