span-reporter span id generation is very slow. #18

[rohitprg]

This is the PR for issue #18
Span-reporter should have an option to use the tracing implementation's span id instead of generating its own.
Span-reporter method SpanBuilderR.findSpanId calls context.baggageItems() twice.

[rohitprg]

Span-reporter uses UUID.randomUUID() to generate the span id, which internally uses SecureRandom and is very slow.

Whereas tracer server like Jaeger uses Java6CompatibleThreadLocalRandom.current().nextLong() which is 10 times faster with no contention(one thread) and around 50 to 100 times faster with contention (multilple threads) compared to UUID.randomUUID().

Instead of creating its own span id, the span-reporter lib can get the span id from the tracing implementation.

Can someone please review the MR? Any feedback would be highly appreciated.

[rohitprg]

Reason behind second commit (Override toString() method)

Calling toString() on the Tracer when using span-reporter does not return useful data.
For example, calling GlobalTracer.get().toString() returns:

"GlobalTracer{io.opentracing.contrib.reporter.TracerR@5eb97ced}"

However, if you do the same thing when directly using the Jaeger tracing you get:

"GlobalTracer{JaegerTracer(version=Java-1.6.0, serviceName=boss, reporter=RemoteReporter(sender=HttpSender(), closeEnqueueTimeout=1000), sampler=ProbabilisticSampler(tags={sampler.type=probabilistic, sampler.param=1.0}), tags={hostname=ffff-dev, jaeger.version=Java-1.6.0, java.user.name=bmhhhh, java.version=11.0.13, boss.signature=0.1.0-SNAPSHOT/f4bc2fc3/fffff/2021-11-17T18:40:34Z, ip=**************}, zipkinSharedRpcSpan=false, expandExceptionLogs=false, useTraceId128Bit=false)}"

io.opentracing.contrib.reporter.TracerR needs to implement toString and return a value that includes the data returned by the wrapped tracer.

[rohitprg]

It has been a while since I opened the merge request. Can someone please review and provide the access to merge?

[rohitprg]

@davidB Can you please review the pull request?

[davidB]

Thanks for the contribution,
I need to go back (refresh) into the API of Opentracing (like spanContext.toSpanId();) and into the code to complete my review.
FYI, I'll not be able to publish a new release (I don't have the permissions to publish)

[davidB]

regarding the impl of findSpanId, referenceId is always "" in this case

[rohitprg]

Thanks for the contribution, I need to go back (refresh) into the API of Opentracing (like spanContext.toSpanId();) and into the code to complete my review. FYI, I'll not be able to publish a new release (I don't have the permissions to publish)

@davidB - Thanks for your reply. Does that mean there wont be any releases/updates on this library going forward?

[sriramraghavan-oracle]

Hello @davidB and @bhs
Can you guys help in closing this fix and help with a new release?
Thanks

[bhs]

It has been an eternity since I worked on this code and I can't even build it anymore, my apologies! My own efforts in the overall space are completely focused on opentelemetry these days (opentracing has been deprecated).

Not the answer you're hoping for, I'm sure, but I wanted to respond at least. :-/