fix(EUM-server): Add conversion-service to EUM server

[aaronweissler]

This change is 

[heiko-holz]

Reviewed 4 of 4 files at r1, all commit messages.
Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @aaronweissler)

---

components/inspectit-ocelot-eum-server/src/main/java/rocks/inspectit/oce/eum/server/exporters/beacon/BeaconHttpExporter.java, line 22 at r1 (raw file):

@Slf4j
@ConditionalOnProperty({"inspectit-eum-server.exporters.beacons.http.enabled"})
@ConditionalOnExpression("NOT new String('${inspectit-eum-server.exporters.beacons.http.enabled}').toUpperCase().equals(T(rocks.inspectit.ocelot.config.model.exporters.ExporterEnabledState).DISABLED.toString())")

why do we need to convert it to toUpperCase()?
Is the value parsed from deprecated configurations disabled, DISABLED (according to BooleanToExporterEnabledStateConverter.java), or false?

We could potentially also try to parse the ExporterEnabledState from the String and checker whether it is .disabled(), i.e. ExporterEnabledState.valueOf(new String('${inspectit-eum-server.exporters.beacons.http.enabled}')).isDisabled()

[aaronweissler]

Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @heiko-holz)

---

components/inspectit-ocelot-eum-server/src/main/java/rocks/inspectit/oce/eum/server/exporters/beacon/BeaconHttpExporter.java, line 22 at r1 (raw file):

Previously, heiko-holz (Heiko Holz) wrote…

why do we need to convert it to toUpperCase()?
Is the value parsed from deprecated configurations disabled, DISABLED (according to BooleanToExporterEnabledStateConverter.java), or false?

We could potentially also try to parse the ExporterEnabledState from the String and checker whether it is .disabled(), i.e. ExporterEnabledState.valueOf(new String('${inspectit-eum-server.exporters.beacons.http.enabled}')).isDisabled()

In both cases the problem is that this evaluation seems to happen really early on the raw input data before Spring uses the converters or even parses it.

The toUpperCase() is then needed, because someone could put in "disabled" as the enabled value which Spring would accept just fine and later parse as ExporterEnabledState.DISABLED, but in the evaluation it is still seen as "disabled" which would lead to the Component being built despite actually being supposed to be disabled. This part is a correction for the new correct way of setting enabled.

The deprecated way would be setting it to false, which is filtered by the ConditionalOnProperty annotation already.

This early evaluation also makes the recommendation impossible unfortunately, because if the old deprecated way is used this would throw an Exception, since it would try to parse "false" or "true" with valueOf, because they will only get converted correctly afterwards.

[heiko-holz]

Reviewable status: complete! all files reviewed, all discussions resolved (waiting on @aaronweissler)

---

components/inspectit-ocelot-eum-server/src/main/java/rocks/inspectit/oce/eum/server/exporters/beacon/BeaconHttpExporter.java, line 22 at r1 (raw file):

Previously, aaronweissler wrote…

In both cases the problem is that this evaluation seems to happen really early on the raw input data before Spring uses the converters or even parses it.

The toUpperCase() is then needed, because someone could put in "disabled" as the enabled value which Spring would accept just fine and later parse as ExporterEnabledState.DISABLED, but in the evaluation it is still seen as "disabled" which would lead to the Component being built despite actually being supposed to be disabled. This part is a correction for the new correct way of setting enabled.

The deprecated way would be setting it to false, which is filtered by the ConditionalOnProperty annotation already.

This early evaluation also makes the recommendation impossible unfortunately, because if the old deprecated way is used this would throw an Exception, since it would try to parse "false" or "true" with valueOf, because they will only get converted correctly afterwards.

thanks for the exhaustive explanation!