nginx shows sometimes upstream_response_time larger than request_time
- systemtap
- openresty must configure with --with-debug --with-dtrace-probes
# start backend-server
cd backend-server
openresty -p `pwd` -c nginx.conf
# start main server
cd ..
openresty -p `pwd` -c nginx.conf
# tracing with systemtap
stap -v -x {main server worker pid} -I tapset stap-scripts/request_time.stp
curl localhost:8080