Hi Steve,

in regards to the JIRA issue you added here, it looks like these metrics
are for MicroProfile component (
https://camel.apache.org/components/latest/microprofile-metrics-component.html),
perhaps you can try with microprofile component instead?

Regards,
Omar

On Sat, Jan 18, 2020 at 12:16 AM Steve Davids <sdav...@gmail.com> wrote:

> I have recently migrated to Camel 3.0 (using Spring Boot) and found that
> there was a change in behavior for our metrics which drives some dashboards
> which shows the number of requests in our primary "process" exchange. The
> old code performed a prometheus query which counted the sum of  an earlier
> stage's CamelMessageHistory_seconds_count and compared it to the count of
> the same metric of the primary "process" stage. Camel 2.x seemed to only
> increment the count upon exit of the stage while Camel 3.x seems to
> increment the count upon starting the stage. This nuance throws off our
> timing graphing and we can't simply choose the "next" component in the
> chain because there is global exception handling that might be triggered
> from the process which wouldn't allow us to watch a single component "next
> in line".
>
> Taking a step back, I would like to find out what the best way to get a
> simple count of "in flight" count into Micrometer/Prometheus. Ideally Camel
> would provide this functionality natively and would allow me to filter on
> the nodeId of the process I'm interested in, but that metric doesn't seem
> to be exposed (though it seems to be exposed for other metrics components
> <https://issues.apache.org/jira/browse/CAMEL-13977>). I also tried to do:
>
> > .doTry()
> >     .toF("micrometer:counter:%s", INFLIGHT_METRIC_NAME)
> >     .process(handler)
> >     .doFinally()
> >         .toF("micrometer:counter:%s?decrement=1", INFLIGHT_METRIC_NAME)
> >     .end()
> >
>
> But with that configuration my global exception handling breaks for the
> `process(handler)` code. Does anyone have suggestions as to how to
> accomplish this or if Camel should natively support this use-case?
>
> Thanks,
>
> -Steve
>

Reply via email to