The metric scope options are used by reporters that do not rely on tags,
to generate a fully-qualified metric name.
The prometheus reporter identifies metrics in a different way (a generic
scope like jobmanager.job.myMetric + a bunch of tags to select a
specific instance) instead, and ignores the metric scope entirely.
On 20/10/2021 16:28, JP MB wrote:
Hi,
Sorry for highlighting this one again.
Can someone provide me some light on the purpose of metric scope
configurations and where can I see the immediate results of changing
these properties?
Regards,
José Brandão
Em ter., 19 de out. de 2021 às 18:19, JP MB
<jose.brandao1...@gmail.com> escreveu:
Hello,
I have been playing with metric scopes and I'm not sure if I
understood them correctly.
https://ci.apache.org/projects/flink/flink-docs-master/docs/ops/metrics/
For instance,
* |metrics.scope.task|
o Default:
<host>.taskmanager.<tm_id>.<job_name>.<task_name>.<subtask_index>
o Applied to all metrics that were scoped to a task.
*What is supposed to happen when if remove the <subtask_index>
scope from there? *
* |metrics.scope.task=|<host>.taskmanager.<tm_id>.<job_name>.<task_name>
I have performed this change but was unable to see any differences
in the metrics that are being exported by the reporter I'm
currently using which is the PrometheusReporter.
*Also, is there any place where I can see the "raw"/pre-reporter
metrics?
*
Regards,
José Brandão
