Platform HTTP component under load won't serve health endpoints anymore

[tstuber]

We are stress-testing a camel container that exposes a HTTP endpoint via platform HTTP. The container calls some legacy backend services - a single process might take some seconds.
However, if we invoke a lot of request (>500), than the container is not anymore able to serve the liveness probe on OpenShift and will be killed.

Potential workarounds:

• Changing liveness probe to run some bash script (but we would like to use the camel-health endpoint)
• Expose the API with Netty on a separate endpoint. This works fine, but this is somehow not that nice.

[tstuber]

See also this discussion on Camel zulip chat.

[jamesnetherton]

A dedicated management interface (running on its own server / port) is coming in the next Quarkus release.