Lag when adding aws xray daemon

[jpresley23-tw]

We saw an increase of about 3 seconds when we started to send transactions to the aws xray daemon from our spring boot application. While debugging with time nc -vz -u localhost 2000 we see there is a 3s duration for the connection when we add the aws xray daemon as a sidecar to our debug pod. We get the same result when we try to run the aws xray daemon as a deployment. We tried to change the port but got the same result. Is there a way to try to use TCP protocol instead of UDP protocol for the aws xray daemon or is there something else we can try to remove the lag we noticed after we added the aws xray daemon?

[wangzlei]

• Launching X-Ray Daemon in docker is fast.
• UDP packets are connectionless it does not bock application.

So, we are not sure the lag would be solved by TCP or Daemon improvement. Could you provide the info of time nc -vz -u localhost 2000 and how do you instrument spring boot?

[jpresley23-tw]

Since posting my issue I went back to using a deployment with a service in front of it for the X-Ray Daemon and I tried changing the port to 3000. I get the same results. The output is

time nc -vz -u aws-xray-service.aws-xray 3000
Connection to aws-xray-service.aws-xray (10.100.47.15) 3000 port [udp/*] succeeded!

real	0m3.016s
user	0m0.002s
sys	0m0.000s

We instrumented spring boot with:
in application.yml

  zipkin:
    internal:
      module:
        xray: zipkin.module.aws.xray.ZipkinXRayStorageModule
    storage:
      xray:
        daemon-address: aws-xray-service.aws-xray:2000

and in build.gradle

implementation 'io.zipkin.aws:zipkin-reporter-xray-udp:0.23.4'