Closed damemi closed 1 year ago
Proposed Plan:
faas.name
to job
, and faas.instance
to instance
.faas.instance
instead of faas.id
, and update metric exporter mapping.For anyone using our current documentation, they would not be broken at any point.
@aabmass do you think that plan makes sense?
SGTM. Do we consider this breaking for the genericTask mapping Mike mentioned? This would apply to users sending to Cloud Monitoring (not GMP). I think it should be ok
It is breaking unless they are using our resource detector, which seems OK. We should recommend people use our resource detectors wherever possible.
I need to update contrib for golang before I can update the this repo
The collector's resource detector needs to be updated. I had forgotten about that.
As of semconv v1.19.0,
faas.id
is removed(/replaced withcloud.resource_id
).So, when we update to semconv v1.19, we'll need to update our usages of
faas.id
tofaas.instance
Sorry it's been a while and I'm confused again. If faas.id
was renamed to cloud.resource_id
, shouldn't we be using that? Or were we incorrectly using faas.id
in the first place?
We were incorrectly using faas.id in the first place. faas.id
was meant to be the full resource name (e.g. projects/my-project/locations/us-central1/functions/my-function)
We can track documentation updates in https://github.com/GoogleCloudPlatform/opentelemetry-cloud-run/issues/6. All other work is done
Our resource detectors and monitored resource mappings currently rely on semconv v1.18, which has
faas.id
, which we use for the Generic Task (Cloud Run) resource.As of semconv v1.19.0,
faas.id
is removed(/replaced withcloud.resource_id
).So, when we update to semconv v1.19, we'll need to update our usages of
faas.id
tofaas.instance
:This attribute is also used in our code samples for GMP-Cloud Run integration: https://github.com/GoogleCloudPlatform/golang-samples/blob/e4eddad4457dc10f0fabb1b6f4c1906bea46aedc/run/custom-metrics/collector/collector-config.yaml#L45, which will need to be updated and the matching docs pages manually redeployed.