[
https://issues.apache.org/jira/browse/FLINK-38154?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18010681#comment-18010681
]
Robert Metzger edited comment on FLINK-38154 at 7/29/25 2:15 PM:
-----------------------------------------------------------------
Thanks for the ticket.
Why are you proposing to scope the `method` under `factory`?
Other metrics reporters make it a top level config:
{code:java}
metrics.reporter.grph.protocol: TCP
{code}
And maybe also going for protocol would be more consistent?
was (Author: rmetzger):
Thanks for the ticket.
Why are you proposing to scope the `method` under `factory`?
Other metrics reporters make it a top level config:
{code:java}
metrics.reporter.grph.protocol: TCP
{code}
> Add HTTP support to OpenTelemetry Reporter
> ------------------------------------------
>
> Key: FLINK-38154
> URL: https://issues.apache.org/jira/browse/FLINK-38154
> Project: Flink
> Issue Type: Improvement
> Components: Runtime / Metrics
> Reporter: Kumar Mallikarjuna
> Priority: Major
>
> *Motivation*
> Currently, Flink's _{{OpenTelemetryMetricReporter}}_ only supports exporting
> metrics via OTLP/gRPC. However, many modern observability platforms, like
> Prometheus, ingest metrics using OTLP/HTTP. This limitation obstructs direct
> integration with such systems.
> While workarounds exist, such as deploying an intermediary OpenTelemetry
> Collector to translate OTLP/gRPC to OTLP/HTTP, they add an extra dependency
> and operational overhead.
>
> *Goal*
> Add OTLP support for HTTP in {_}{{OpenTelemetryMetricReporter}}{_}. The user
> should be able to choose the protocol using the following configuration
> {code:java}
> metrics.reporter.otel.factory.method: {grpc, http}{code}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
