let's first test with CI, then hush-house
since we seem to be using a unix socket for dogstatsd communication with the datadog agent, it makes sense to use a socket for APM as well. i think we'll need to upgrade the datadog chart in order for this configuration to be available -- I think the first version of the datadog chart that allowed this was 2.3.5.
Once that part is done i think it's enough to have concourse send its jaeger spans to the APM agent... it just occurred to me that this may not be possible because concourse may not be able to send HTTP messages to a unix socket.
let's first test with CI, then hush-house since we seem to be using a unix socket for dogstatsd communication with the datadog agent, it makes sense to use a socket for APM as well. i think we'll need to upgrade the datadog chart in order for this configuration to be available -- I think the first version of the datadog chart that allowed this was 2.3.5.
Once that part is done i think it's enough to have concourse send its jaeger spans to the APM agent... it just occurred to me that this may not be possible because concourse may not be able to send HTTP messages to a unix socket.
In any case, these docs ought to help: https://docs.datadoghq.com/agent/kubernetes/apm/?tab=helm -- they suggest using the downward API to expose the host IP and connecting to the trace via that.