Closed Raynos closed 8 years ago
@Raynos Thanks a lot for the analysis. lgtm. I will use this branch as a starting point for finishing up the opentracing diff that Rooz started.
All files effected are in the benchmarks/
folder so I'm going to go ahead and land this.
The Makefile is not configured to work out of the box.
There was also one bug in the benchmark bootstrapper that caused the tracing benchmark to fail
I've also turned on 100% trace reporting instead of 1% for the tracing benchmark...
So currently when benchmarking wrt to tracing:
make take
will benchmark p2p, with span creation and noop-reportermake take_relay
will benchmark with relay, with span creation & noop-reportermake trace
will benchmark with relay, with span creation & 100% reporterThe four modes do the following on my laptop,
make take
, hand-edited to have trace: false, aka no span creationmake take
, normal, span creation, noop reportermake take_relay
, normal, span creation, relay, noop reportermake take_trace
, normal, span creation, relay, 100% tcollector reporterWe currently do not have a
make take_no_trace
so you have to hand edit and s/trace: true/trace: false/.r: @rf @oibe