Prometheus example is panic (exporting histogram panic)

[youngbupark]

SDK version

>= 0.5.0

No issue with 0.4.3.

Repro step

1. Run prometheus example - https://github.com/open-telemetry/opentelemetry-go/blob/master/example/prometheus/main.go

2. Browse localhost:2222 to see the metrics after 1 minute

Expected Result

It should show the list of metrics on the browser without any panic.

Actual Result

panic: runtime error: index out of range [-1]

goroutine 33 [running]:
go.opentelemetry.io/otel/exporters/metric/prometheus.(*collector).exportHistogram(0xc000116018, 0xc00013a720, 0x1f88440, 0xc0003d4dc0, 0xc0003d4d01, 0xc000134230, 0xc00012a3a0, 0x1, 0x1, 0x0, ...)
	/Users/youngp/go/pkg/mod/go.opentelemetry.io/otel/exporters/metric/prometheus@v0.5.0/prometheus.go:326 +0x870
go.opentelemetry.io/otel/exporters/metric/prometheus.(*collector).Collect.func1(0xc0001320f8, 0xc000144228, 0x16ca860, 0xc0003d4dc0, 0x0, 0x0)
	/Users/youngp/go/pkg/mod/go.opentelemetry.io/otel/exporters/metric/prometheus@v0.5.0/prometheus.go:218 +0x304
go.opentelemetry.io/otel/sdk/metric/integrator/simple.batchMap.ForEach(0xc000126180, 0xc000115fe0, 0x0, 0x0)
	/Users/youngp/go/pkg/mod/go.opentelemetry.io/otel@v0.5.0/sdk/metric/integrator/simple/simple.go:110 +0x19c
go.opentelemetry.io/otel/sdk/metric/controller/push.syncCheckpointSet.ForEach(0xc00013e008, 0x16c47e0, 0xc000126180, 0xc000115fe0, 0x0, 0x0)
	/Users/youngp/go/pkg/mod/go.opentelemetry.io/otel@v0.5.0/sdk/metric/controller/push/push.go:208 +0xa9
go.opentelemetry.io/otel/exporters/metric/prometheus.(*collector).Collect(0xc000116018, 0xc00013a720)
	/Users/youngp/go/pkg/mod/go.opentelemetry.io/otel/exporters/metric/prometheus@v0.5.0/prometheus.go:211 +0xcc
github.com/prometheus/client_golang/prometheus.(*Registry).Gather.func1()
	/Users/youngp/go/pkg/mod/github.com/prometheus/client_golang@v1.5.0/prometheus/registry.go:445 +0x21e
created by github.com/prometheus/client_golang/prometheus.(*Registry).Gather
	/Users/youngp/go/pkg/mod/github.com/prometheus/client_golang@v1.5.0/prometheus/registry.go:454 +0x803

This panic caused because buckets.Counts was nil

opentelemetry-go/exporters/metric/prometheus/prometheus.go

Line 326 in 6bc14ff

totalCount += buckets.Counts[len(buckets.Counts)-1].AsUint64()

Adding DefaultHistogramBoundaries setting to prometheus config did not work.

[surajkjai]

I modified the last sleep statement for a 5 second sleep to avoid the reported crash. A curl of the endpoint localhost:2222 does not produce any output.

~/opentelemetry-go/example/prometheus$ curl -vvv localhost:2222/

• Trying 127.0.0.1...
• TCP_NODELAY set
• Connected to localhost (127.0.0.1) port 2222 (#0)

GET / HTTP/1.1
Host: localhost:2222
User-Agent: curl/7.58.0
Accept: /

< HTTP/1.1 200 OK
< Content-Type: text/plain; version=0.0.4; charset=utf-8
< Date: Mon, 18 May 2020 07:03:45 GMT
< Content-Length: 0
<

• Connection #0 to host localhost left intact

[jmacd]

I was able to reproduce this, though. See #736.