1
0
mirror of https://github.com/open-telemetry/opentelemetry-go.git synced 2024-12-12 10:04:29 +02:00
opentelemetry-go/exporters/trace/jaeger/jaeger.go
Johannes Liebermann c3c4273ecc
Add RO/RW span interfaces (#1360)
* Store span data directly in the span

- Nesting only some of a span's data in a `data` field (with the rest
  of the data living direclty in the `span` struct) is confusing.
- export.SpanData is meant to be an immutable *snapshot* of a span,
  not the "authoritative" state of the span.
- Refactor attributesMap.toSpanData into toKeyValue and make it
  return a []label.KeyValue which is clearer than modifying a struct
  passed to the function.
- Read droppedCount from the attributesMap as a separate operation
  instead of setting it from within attributesMap.toSpanData.
- Set a span's end time in the span itself rather than in the
  SpanData to allow reading the span's end time after a span has
  ended.
- Set a span's end time as soon as possible within span.End so that
  we don't influence the span's end time with operations such as
  fetching span processors and generating span data.
- Remove error handling for uninitialized spans. This check seems to
  be necessary only because we used to have an *export.SpanData field
  which could be nil. Now that we no longer have this field I think we
  can safely remove the check. The error isn't used anywhere else so
  remove it, too.

* Store parent as trace.SpanContext

The spec requires that the parent field of a Span be a Span, a
SpanContext or null.

Rather than extracting the parent's span ID from the trace.SpanContext
which we get from the tracer, store the trace.SpanContext as is and
explicitly extract the parent's span ID where necessary.

* Add ReadOnlySpan interface

Use this interface instead of export.SpanData in places where reading
information from a span is necessary. Use export.SpanData only when
exporting spans.

* Add ReadWriteSpan interface

Use this interface instead of export.SpanData in places where it is
necessary to read information from a span and write to it at the same
time.

* Rename export.SpanData to SpanSnapshot

SpanSnapshot represents the nature of this type as well as its
intended use more accurately.

Clarify the purpose of SpanSnapshot in the docs and emphasize what
should and should not be done with it.

* Rephrase attributesMap doc comment

"refreshes" is wrong for plural ("updates").

* Refactor span.End()

- Improve accuracy of span duration. Record span end time ASAP. We
  want to measure a user operation as accurately as possible, which
  means we want to mark the end time of a span as soon as possible
  after span.End() is called. Any operations we do inside span.End()
  before storing the end time affect the total duration of the span,
  and although these operations are rather fast at the moment they
  still seem to affect the duration of the span by "artificially"
  adding time between the start and end timestamps. This is relevant
  only in cases where the end time isn't explicitly specified.
- Remove redundant idempotence check. Now that IsRecording() is based
  on the value of span.endTime, IsRecording() will always return
  false after span.End() had been called because span.endTime won't
  be zero. This means we no longer need span.endOnce.
- Improve TestEndSpanTwice so that it also ensures subsequent calls
  to span.End() don't modify the span's end time.

* Update changelog

Co-authored-by: Tyler Yahn <codingalias@gmail.com>
Co-authored-by: Tyler Yahn <MrAlias@users.noreply.github.com>
2020-12-10 21:15:44 -08:00

449 lines
12 KiB
Go

// Copyright The OpenTelemetry Authors
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
package jaeger // import "go.opentelemetry.io/otel/exporters/trace/jaeger"
import (
"context"
"encoding/binary"
"fmt"
"sync"
"google.golang.org/api/support/bundler"
"go.opentelemetry.io/otel"
"go.opentelemetry.io/otel/codes"
gen "go.opentelemetry.io/otel/exporters/trace/jaeger/internal/gen-go/jaeger"
"go.opentelemetry.io/otel/label"
export "go.opentelemetry.io/otel/sdk/export/trace"
sdktrace "go.opentelemetry.io/otel/sdk/trace"
"go.opentelemetry.io/otel/trace"
)
const (
defaultServiceName = "OpenTelemetry"
keyInstrumentationLibraryName = "otel.instrumentation_library.name"
keyInstrumentationLibraryVersion = "otel.instrumentation_library.version"
)
type Option func(*options)
// options are the options to be used when initializing a Jaeger export.
type options struct {
// Process contains the information about the exporting process.
Process Process
// BufferMaxCount defines the total number of traces that can be buffered in memory
BufferMaxCount int
// BatchMaxCount defines the maximum number of spans sent in one batch
BatchMaxCount int
Config *sdktrace.Config
Disabled bool
}
// WithProcess sets the process with the information about the exporting process.
func WithProcess(process Process) Option {
return func(o *options) {
o.Process = process
}
}
// WithBufferMaxCount defines the total number of traces that can be buffered in memory
func WithBufferMaxCount(bufferMaxCount int) Option {
return func(o *options) {
o.BufferMaxCount = bufferMaxCount
}
}
// WithBatchMaxCount defines the maximum number of spans in one batch
func WithBatchMaxCount(batchMaxCount int) Option {
return func(o *options) {
o.BatchMaxCount = batchMaxCount
}
}
// WithSDK sets the SDK config for the exporter pipeline.
func WithSDK(config *sdktrace.Config) Option {
return func(o *options) {
o.Config = config
}
}
// WithDisabled option will cause pipeline methods to use
// a no-op provider
func WithDisabled(disabled bool) Option {
return func(o *options) {
o.Disabled = disabled
}
}
// NewRawExporter returns an OTel Exporter implementation that exports the
// collected spans to Jaeger.
//
// It will IGNORE Disabled option.
func NewRawExporter(endpointOption EndpointOption, opts ...Option) (*Exporter, error) {
uploader, err := endpointOption()
if err != nil {
return nil, err
}
o := options{}
opts = append(opts, WithProcessFromEnv())
for _, opt := range opts {
opt(&o)
}
service := o.Process.ServiceName
if service == "" {
service = defaultServiceName
}
tags := make([]*gen.Tag, 0, len(o.Process.Tags))
for _, tag := range o.Process.Tags {
t := keyValueToTag(tag)
if t != nil {
tags = append(tags, t)
}
}
e := &Exporter{
uploader: uploader,
process: &gen.Process{
ServiceName: service,
Tags: tags,
},
o: o,
}
bundler := bundler.NewBundler((*gen.Span)(nil), func(bundle interface{}) {
if err := e.upload(bundle.([]*gen.Span)); err != nil {
otel.Handle(err)
}
})
// Set BufferedByteLimit with the total number of spans that are permissible to be held in memory.
// This needs to be done since the size of messages is always set to 1. Failing to set this would allow
// 1G messages to be held in memory since that is the default value of BufferedByteLimit.
if o.BufferMaxCount != 0 {
bundler.BufferedByteLimit = o.BufferMaxCount
}
// The default value bundler uses is 10, increase to send larger batches
if o.BatchMaxCount != 0 {
bundler.BundleCountThreshold = o.BatchMaxCount
}
e.bundler = bundler
return e, nil
}
// NewExportPipeline sets up a complete export pipeline
// with the recommended setup for trace provider
func NewExportPipeline(endpointOption EndpointOption, opts ...Option) (trace.TracerProvider, func(), error) {
o := options{}
opts = append(opts, WithDisabledFromEnv())
for _, opt := range opts {
opt(&o)
}
if o.Disabled {
return trace.NewNoopTracerProvider(), func() {}, nil
}
exporter, err := NewRawExporter(endpointOption, opts...)
if err != nil {
return nil, nil, err
}
pOpts := []sdktrace.TracerProviderOption{sdktrace.WithSyncer(exporter)}
if exporter.o.Config != nil {
pOpts = append(pOpts, sdktrace.WithConfig(*exporter.o.Config))
}
tp := sdktrace.NewTracerProvider(pOpts...)
return tp, exporter.Flush, nil
}
// InstallNewPipeline instantiates a NewExportPipeline with the
// recommended configuration and registers it globally.
func InstallNewPipeline(endpointOption EndpointOption, opts ...Option) (func(), error) {
tp, flushFn, err := NewExportPipeline(endpointOption, opts...)
if err != nil {
return nil, err
}
otel.SetTracerProvider(tp)
return flushFn, nil
}
// Process contains the information exported to jaeger about the source
// of the trace data.
type Process struct {
// ServiceName is the Jaeger service name.
ServiceName string
// Tags are added to Jaeger Process exports
Tags []label.KeyValue
}
// Exporter is an implementation of an OTel SpanSyncer that uploads spans to
// Jaeger.
type Exporter struct {
process *gen.Process
bundler *bundler.Bundler
uploader batchUploader
o options
stoppedMu sync.RWMutex
stopped bool
}
var _ export.SpanExporter = (*Exporter)(nil)
// ExportSpans exports SpanSnapshots to Jaeger.
func (e *Exporter) ExportSpans(ctx context.Context, ss []*export.SpanSnapshot) error {
e.stoppedMu.RLock()
stopped := e.stopped
e.stoppedMu.RUnlock()
if stopped {
return nil
}
for _, span := range ss {
// TODO(jbd): Handle oversized bundlers.
err := e.bundler.Add(spanSnapshotToThrift(span), 1)
if err != nil {
return fmt.Errorf("failed to bundle %q: %w", span.Name, err)
}
}
return nil
}
// flush is used to wrap the bundler's Flush method for testing.
var flush = func(e *Exporter) {
e.bundler.Flush()
}
// Shutdown stops the exporter flushing any pending exports.
func (e *Exporter) Shutdown(ctx context.Context) error {
e.stoppedMu.Lock()
e.stopped = true
e.stoppedMu.Unlock()
done := make(chan struct{}, 1)
// Shadow so if the goroutine is leaked in testing it doesn't cause a race
// condition when the file level var is reset.
go func(FlushFunc func(*Exporter)) {
// The OpenTelemetry specification is explicit in not having this
// method block so the preference here is to orphan this goroutine if
// the context is canceled or times out while this flushing is
// occurring. This is a consequence of the bundler Flush method not
// supporting a context.
FlushFunc(e)
done <- struct{}{}
}(flush)
select {
case <-ctx.Done():
return ctx.Err()
case <-done:
}
return nil
}
func spanSnapshotToThrift(ss *export.SpanSnapshot) *gen.Span {
tags := make([]*gen.Tag, 0, len(ss.Attributes))
for _, kv := range ss.Attributes {
tag := keyValueToTag(kv)
if tag != nil {
tags = append(tags, tag)
}
}
// TODO (jmacd): OTel has a broad "last value wins"
// semantic. Should resources be appended before span
// attributes, above, to allow span attributes to
// overwrite resource attributes?
if ss.Resource != nil {
for iter := ss.Resource.Iter(); iter.Next(); {
if tag := keyValueToTag(iter.Attribute()); tag != nil {
tags = append(tags, tag)
}
}
}
if il := ss.InstrumentationLibrary; il.Name != "" {
tags = append(tags, getStringTag(keyInstrumentationLibraryName, il.Name))
if il.Version != "" {
tags = append(tags, getStringTag(keyInstrumentationLibraryVersion, il.Version))
}
}
tags = append(tags,
getInt64Tag("status.code", int64(ss.StatusCode)),
getStringTag("status.message", ss.StatusMessage),
getStringTag("span.kind", ss.SpanKind.String()),
)
// Ensure that if Status.Code is not OK, that we set the "error" tag on the Jaeger span.
// See Issue https://github.com/census-instrumentation/opencensus-go/issues/1041
if ss.StatusCode != codes.Ok && ss.StatusCode != codes.Unset {
tags = append(tags, getBoolTag("error", true))
}
var logs []*gen.Log
for _, a := range ss.MessageEvents {
fields := make([]*gen.Tag, 0, len(a.Attributes))
for _, kv := range a.Attributes {
tag := keyValueToTag(kv)
if tag != nil {
fields = append(fields, tag)
}
}
fields = append(fields, getStringTag("name", a.Name))
logs = append(logs, &gen.Log{
Timestamp: a.Time.UnixNano() / 1000,
Fields: fields,
})
}
var refs []*gen.SpanRef
for _, link := range ss.Links {
refs = append(refs, &gen.SpanRef{
TraceIdHigh: int64(binary.BigEndian.Uint64(link.TraceID[0:8])),
TraceIdLow: int64(binary.BigEndian.Uint64(link.TraceID[8:16])),
SpanId: int64(binary.BigEndian.Uint64(link.SpanID[:])),
// TODO(paivagustavo): properly set the reference type when specs are defined
// see https://github.com/open-telemetry/opentelemetry-specification/issues/65
RefType: gen.SpanRefType_CHILD_OF,
})
}
return &gen.Span{
TraceIdHigh: int64(binary.BigEndian.Uint64(ss.SpanContext.TraceID[0:8])),
TraceIdLow: int64(binary.BigEndian.Uint64(ss.SpanContext.TraceID[8:16])),
SpanId: int64(binary.BigEndian.Uint64(ss.SpanContext.SpanID[:])),
ParentSpanId: int64(binary.BigEndian.Uint64(ss.ParentSpanID[:])),
OperationName: ss.Name, // TODO: if span kind is added then add prefix "Sent"/"Recv"
Flags: int32(ss.SpanContext.TraceFlags),
StartTime: ss.StartTime.UnixNano() / 1000,
Duration: ss.EndTime.Sub(ss.StartTime).Nanoseconds() / 1000,
Tags: tags,
Logs: logs,
References: refs,
}
}
func keyValueToTag(keyValue label.KeyValue) *gen.Tag {
var tag *gen.Tag
switch keyValue.Value.Type() {
case label.STRING:
s := keyValue.Value.AsString()
tag = &gen.Tag{
Key: string(keyValue.Key),
VStr: &s,
VType: gen.TagType_STRING,
}
case label.BOOL:
b := keyValue.Value.AsBool()
tag = &gen.Tag{
Key: string(keyValue.Key),
VBool: &b,
VType: gen.TagType_BOOL,
}
case label.INT32:
i := int64(keyValue.Value.AsInt32())
tag = &gen.Tag{
Key: string(keyValue.Key),
VLong: &i,
VType: gen.TagType_LONG,
}
case label.INT64:
i := keyValue.Value.AsInt64()
tag = &gen.Tag{
Key: string(keyValue.Key),
VLong: &i,
VType: gen.TagType_LONG,
}
case label.UINT32:
i := int64(keyValue.Value.AsUint32())
tag = &gen.Tag{
Key: string(keyValue.Key),
VLong: &i,
VType: gen.TagType_LONG,
}
case label.UINT64:
// we'll ignore the value if it overflows
if i := int64(keyValue.Value.AsUint64()); i >= 0 {
tag = &gen.Tag{
Key: string(keyValue.Key),
VLong: &i,
VType: gen.TagType_LONG,
}
}
case label.FLOAT32:
f := float64(keyValue.Value.AsFloat32())
tag = &gen.Tag{
Key: string(keyValue.Key),
VDouble: &f,
VType: gen.TagType_DOUBLE,
}
case label.FLOAT64:
f := keyValue.Value.AsFloat64()
tag = &gen.Tag{
Key: string(keyValue.Key),
VDouble: &f,
VType: gen.TagType_DOUBLE,
}
}
return tag
}
func getInt64Tag(k string, i int64) *gen.Tag {
return &gen.Tag{
Key: k,
VLong: &i,
VType: gen.TagType_LONG,
}
}
func getStringTag(k, s string) *gen.Tag {
return &gen.Tag{
Key: k,
VStr: &s,
VType: gen.TagType_STRING,
}
}
func getBoolTag(k string, b bool) *gen.Tag {
return &gen.Tag{
Key: k,
VBool: &b,
VType: gen.TagType_BOOL,
}
}
// Flush waits for exported trace spans to be uploaded.
//
// This is useful if your program is ending and you do not want to lose recent spans.
func (e *Exporter) Flush() {
flush(e)
}
func (e *Exporter) upload(spans []*gen.Span) error {
batch := &gen.Batch{
Spans: spans,
Process: e.process,
}
return e.uploader.upload(batch)
}