topic Re: otel operator endpoint invalid - Auto instrumentation broken? in Splunk Observability Cloud

otel operator endpoint invalid - Auto instrumentation broken?

nickhills_encc — Thu, 31 Oct 2024 11:16:06 GMT

On a new fresh deployment of O11y, we are following the guide from the setup wizard and running the following helm install command

helm install splunk-otel-collector --set="cloudProvider=aws,distribution=eks,splunkObservability.accessToken=xxxx-xxxxx,clusterName=eks-test-cluster,splunkObservability.realm=eu2,gateway.enabled=true,splunkPlatform.endpoint=https://my-splunk-cloud-hec-input,splunkPlatform.token=my-token,splunkObservability.profilingEnabled=true,environment=test,operator.enabled=true,agent.discovery.enabled=true" splunk-otel-collector-chart/splunk-otel-collector

However this fails with the following error:

Error: INSTALLATION FAILED: template: splunk-otel-collector/templates/operator/instrumentation.yaml:2:4: executing "splunk-otel-collector/templates/operator/instrumentation.yaml" at <include "splunk-otel-collector.operator.validation-rules" .>: error calling include: template: splunk-otel-collector/templates/operator/_helpers.tpl:17:13: executing "splunk-otel-collector.operator.validation-rules" at <.Values.instrumentation.exporter.endpoint>: nil pointer evaluating interface {}.endpoint

This seems to be because _helpers.tpl is expecting a value for instrumentation.exporter.endpoint however the value according to the chart (and the documentation) is instrumentation.endpoint

https://github.com/signalfx/splunk-otel-collector-chart/blob/main/helm-charts/splunk-otel-collector/templates/operator/_helpers.tpl
Line 13 is where it is mentioned.

We have tried providing instrumentation.exporter.endpoint as an additional parameter - but instead get the error:

Values don't meet the specifications of the schema(s) in the following chart(s):
splunk-otel-collector:
- instrumentation: Additional property exporter is not allowed

(Which is true - instrumentation.exporter.endpoint is not defined in here: https://github.com/signalfx/splunk-otel-collector-chart/blob/main/helm-charts/splunk-otel-collector/templates/operator/instrumentation.yaml )
line 20

We also get the same error if we provide a complete values.yaml file with both formats of the instrumentation endpoint defined.

It looks like _helpers.tpl was edited to include this endpoint specification about a month ago, so surely we can not be the first people to be tripped up by this?

Is there anything else I can try or do we need to wait for the operator to be fixed?

Re: otel operator endpoint invalid - Auto instrumentation broken?

nickhills_encc — Thu, 31 Oct 2024 11:17:37 GMT

Just to follow up that if we disable the operator, the deployment is successful, but we have no APM.

This issue specifically seems to relate to the operator and the APM instrumentation

Re: otel operator endpoint invalid - Auto instrumentation broken?

hdjlassi — Thu, 31 Oct 2024 14:09:17 GMT

Welcome to Splunk O11y Cloud Nickhills !

From your first message, I see that you are deploying the collector as Gateway (on the helm install command, parameter gateway.enabled is set to true). Can you confirm that you need to setup OpenTelemetry as gateway? If not, please try to re-install as agent and let us know how it goes.

Regards,

Houssem

Re: otel operator endpoint invalid - Auto instrumentation broken?

nickhills_encc — Thu, 31 Oct 2024 14:41:23 GMT

helm install -n splunk --create-namespace splunk-otel-collector --set="cloudProvider=aws,distribution=eks,splunkObservability.accessToken=xxxxxx,clusterName=eks-uk-test,splunkObservability.realm=eu2,gateway.enabled=false,splunkPlatform.endpoint=xxxxxxx,splunkPlatform.token=xxxxx,splunkObservability.profilingEnabled=true,environment=test,operator.enabled=true,agent.discovery.enabled=true" splunk-otel-collector-chart/splunk-otel-collector

Still gives:

Error: INSTALLATION FAILED: template: splunk-otel-collector/templates/operator/instrumentation.yaml:2:4: executing "splunk-otel-collector/templates/operator/instrumentation.yaml" at <include "splunk-otel-collector.operator.validation-rules" .>: error calling include: template: splunk-otel-collector/templates/operator/_helpers.tpl:17:13: executing "splunk-otel-collector.operator.validation-rules" at <.Values.instrumentation.exporter.endpoint>: nil pointer evaluating interface {}.endpoint

Re: otel operator endpoint invalid - Auto instrumentation broken?

ramkumarvasu — Thu, 14 Nov 2024 09:07:42 GMT

Yes, same issue when I tried. Only when disabling the operator flag, the deployment goes through

Re: otel operator endpoint invalid - Auto instrumentation broken?

bishida — Thu, 14 Nov 2024 16:53:07 GMT

Hi,

This error is seems to occur with older versions of helm. Can you please confirm your version of helm and see if it's possible to update to a current version?

Re: otel operator endpoint invalid - Auto instrumentation broken?

ramkumarvasu — Fri, 15 Nov 2024 09:02:44 GMT

Helm version.BuildInfo{Version:"v3.14.2", GitCommit:"", GitTreeState:"clean", GoVersion:"go1.22.7"}

I used helm from Azure cloudshell and also tried GCP cloudshell. Both had similar issue. Do I need to try installing kubectl and helm locally and try again?

Re: otel operator endpoint invalid - Auto instrumentation broken?

nickhills_encc — Fri, 15 Nov 2024 09:12:20 GMT

We did eventually resolve this, however it took multiple steps.

Indeed, we were using an old version of helm, updating to 3.16 did allow us to make further progress, however that moved the issue onto a compatibility/dependency issue with prometheus (specifically prometheus-operator)

Switching to the latest otel chart (--version 0.112.0) was one step closer - however there was a breaking change in the values.yaml between 110-112 which meant we needed to rewrite our local values file.

Long story short: helm 3.16 + collector & values 0.112.0 worked for us.

Re: otel operator endpoint invalid - Auto instrumentation broken?

ramkumarvasu — Fri, 15 Nov 2024 09:39:13 GMT

I also tried with the latest version 3.16.3 and it is still the same issue

Re: otel operator endpoint invalid - Auto instrumentation broken?

ramkumarvasu — Sat, 16 Nov 2024 03:31:03 GMT

Can i know what were the changes down on values file? Otel chart I was able to get in the Github project