Implemented a second service for the collector

[jpkrohling]

Fixes #264

Signed-off-by: Juraci Paixão Kröhling [email protected]

[jpkrohling]

This change is 

[jpkrohling]

E2E tests are passing:

Running Smoke end-to-end tests...
ok  	github.com/jaegertracing/jaeger-operator/test/e2e	166.810s
Creating namespace default
Running Cassandra end-to-end tests...
ok  	github.com/jaegertracing/jaeger-operator/test/e2e	116.122s
Running Elasticsearch end-to-end tests...
ok  	github.com/jaegertracing/jaeger-operator/test/e2e	106.595s

[codecov]

Codecov Report

Merging #339 into master will increase coverage by 0.04%.
The diff coverage is 100%.

@@            Coverage Diff            @@
##           master    #339      +/-   ##
=========================================
+ Coverage   89.75%   89.8%   +0.04%     
=========================================
  Files          64      64              
  Lines        2997    3011      +14     
=========================================
+ Hits         2690    2704      +14     
  Misses        207     207              
  Partials      100     100

Impacted Files	Coverage Δ
pkg/inventory/service.go	87.5% <100%> (+0.83%)	⬆️
pkg/service/collector.go	100% <100%> (ø)	⬆️
pkg/inject/sidecar.go	100% <100%> (ø)	⬆️
pkg/deployment/all-in-one.go	100% <100%> (ø)	⬆️
pkg/deployment/collector.go	100% <100%> (ø)	⬆️
pkg/service/query.go	100% <100%> (ø)	⬆️
pkg/deployment/agent.go	100% <100%> (ø)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 0772935...cec8456. Read the comment docs.

[objectiser]

nit: This could take the output from GetNameForCollectorService and append -headless. Only minor though feel free to ignore.

[jpkrohling]

I manually tested the load balancing for the communication between the agent => collector and it's still working after this PR.

https://gist.github.com/jpkrohling/0de49a9c23c24f4db07427aee7206a0e

When setting the JAEGER_ENDPOINT, pointing to the cluster IP address, each tracer instance sends spans to the same collector, which can probably be explained by OkHttp caching the connection.

Here's the YAML definition for the test application:

apiVersion: apps/v1
kind: Deployment
metadata:
  name: myapp
spec:
  selector:
    matchLabels:
      app: myapp
  template:
    metadata:
      labels:
        app: myapp
    spec:
      containers:
      - name: myapp
        image: jaegertracing/vertx-create-span:operator-e2e-tests
        env:
        - name: JAEGER_ENDPOINT
          value: "http://simple-prod-collector:14268/api/traces"

[jpkrohling]

@adinunzio84, would you take a look at this PR and see if it solves the problem you reported?

[objectiser]

@jpkrohling Could you scale up the test app, and see whether the use of JAEGER_ENDPOINT also load balances?

[jpkrohling]

It does. Doesn't look as even as when using the agent, but it does balance. Note that the output here is skewed, as it still has data from the first run:

$ curl localhost:14268/metrics 2>/dev/null | grep jaeger_collector_spans_received_total 
jaeger_collector_spans_received_total{debug="false",format="jaeger",svc="inventory"} 800
jaeger_collector_spans_received_total{debug="false",format="jaeger",svc="order"} 500

$ curl localhost:14268/metrics 2>/dev/null | grep jaeger_collector_spans_received_total 
jaeger_collector_spans_received_total{debug="false",format="jaeger",svc="inventory"} 2000
jaeger_collector_spans_received_total{debug="false",format="jaeger",svc="order"} 1000

$ curl localhost:14268/metrics 2>/dev/null | grep jaeger_collector_spans_received_total 
jaeger_collector_spans_received_total{debug="false",format="jaeger",svc="inventory"} 400
jaeger_collector_spans_received_total{debug="false",format="jaeger",svc="order"} 2500

[objectiser]

@jpkrohling ok great!

[jpkrohling]

I'm merging this, but let me know if there's anything missing, @adinunzio84.