1
0
mirror of https://github.com/open-telemetry/opentelemetry-go.git synced 2025-01-26 03:52:03 +02:00

288 lines
7.8 KiB
Go
Raw Normal View History

// 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 global // import "go.opentelemetry.io/otel/internal/metric/global"
import (
"context"
"sync"
"sync/atomic"
"unsafe"
"go.opentelemetry.io/otel/attribute"
Separate InstrumentationLibrary from metric.Descriptor (#2197) * factor instrumentation library out of the instrument descriptor * SDK tests pass * checkpoint work * otlp and opencensus tests passing * prometheus * tests pass, working on lint * lint applied: MetricReader->Reader * comments * Changelog * Apply suggestions from code review Co-authored-by: alrex <alrex.boten@gmail.com> * remove an interdependency * fix build * re-indent one * Apply suggestions from code review Co-authored-by: Tyler Yahn <MrAlias@users.noreply.github.com> * Lint&feedback * update after rename * comment fix * style fix for meter options * remove libraryReader, let Controller implement the reader API directly * rename 'impl' field to 'provider' * remove a type assertion * move metric/registry into internal; move registry.MeterProvider into metric controller * add test for controller registry function * CheckpointSet->Reader everywhere * lint * remove two unnecessary accessor methods; Controller implements MeterProvider and InstrumentationLibraryReader directly, no need to get these * use a sync.Map * ensure the initOnce is always called; handle multiple errors * Apply suggestions from code review Co-authored-by: Anthony Mirabella <a9@aneurysm9.com> * cleanup locking in metrictest * Revert "ensure the initOnce is always called; handle multiple errors" This reverts commit 3356eb5ed0c288ac3edcc2bc2e853aecda7f29b3. * Revert "use a sync.Map" This reverts commit ea7bc599bd3a24c4acb4cd9facd13f08cd702237. * restore the TODO about sync.Map Co-authored-by: alrex <alrex.boten@gmail.com> Co-authored-by: Tyler Yahn <MrAlias@users.noreply.github.com> Co-authored-by: Anthony Mirabella <a9@aneurysm9.com>
2021-09-27 08:51:47 -07:00
"go.opentelemetry.io/otel/internal/metric/registry"
"go.opentelemetry.io/otel/metric"
"go.opentelemetry.io/otel/metric/number"
"go.opentelemetry.io/otel/metric/sdkapi"
)
// This file contains the forwarding implementation of MeterProvider used as
// the default global instance. Metric events using instruments provided by
// this implementation are no-ops until the first Meter implementation is set
// as the global provider.
//
// The implementation here uses Mutexes to maintain a list of active Meters in
// the MeterProvider and Instruments in each Meter, under the assumption that
// these interfaces are not performance-critical.
//
// We have the invariant that setDelegate() will be called before a new
// MeterProvider implementation is registered as the global provider. Mutexes
// in the MeterProvider and Meters ensure that each instrument has a delegate
// before the global provider is set.
//
// Metric uniqueness checking is implemented by calling the exported
// methods of the api/metric/registry package.
type meterKey struct {
Separate InstrumentationLibrary from metric.Descriptor (#2197) * factor instrumentation library out of the instrument descriptor * SDK tests pass * checkpoint work * otlp and opencensus tests passing * prometheus * tests pass, working on lint * lint applied: MetricReader->Reader * comments * Changelog * Apply suggestions from code review Co-authored-by: alrex <alrex.boten@gmail.com> * remove an interdependency * fix build * re-indent one * Apply suggestions from code review Co-authored-by: Tyler Yahn <MrAlias@users.noreply.github.com> * Lint&feedback * update after rename * comment fix * style fix for meter options * remove libraryReader, let Controller implement the reader API directly * rename 'impl' field to 'provider' * remove a type assertion * move metric/registry into internal; move registry.MeterProvider into metric controller * add test for controller registry function * CheckpointSet->Reader everywhere * lint * remove two unnecessary accessor methods; Controller implements MeterProvider and InstrumentationLibraryReader directly, no need to get these * use a sync.Map * ensure the initOnce is always called; handle multiple errors * Apply suggestions from code review Co-authored-by: Anthony Mirabella <a9@aneurysm9.com> * cleanup locking in metrictest * Revert "ensure the initOnce is always called; handle multiple errors" This reverts commit 3356eb5ed0c288ac3edcc2bc2e853aecda7f29b3. * Revert "use a sync.Map" This reverts commit ea7bc599bd3a24c4acb4cd9facd13f08cd702237. * restore the TODO about sync.Map Co-authored-by: alrex <alrex.boten@gmail.com> Co-authored-by: Tyler Yahn <MrAlias@users.noreply.github.com> Co-authored-by: Anthony Mirabella <a9@aneurysm9.com>
2021-09-27 08:51:47 -07:00
InstrumentationName string
InstrumentationVersion string
SchemaURL string
}
type meterProvider struct {
delegate metric.MeterProvider
Add observer metric (#474) * wip: observers * wip: float observers * fix copy pasta * wip: rework observers in sdk * small fix in global meter * wip: aggregators and selectors * wip: monotonicity option for observers * some refactor * wip: docs needs more package docs (especially for api/metric and sdk/metric) * fix ci * Fix copy-pasta in docs Co-Authored-By: Mauricio Vásquez <mauricio@kinvolk.io> * recycle unused recorders in observers if a recorder for a labelset is unused for a second collection cycle in a row, drop it * unregister * thread-safe set callback * Fix docs * Revert "wip: aggregators and selectors" This reverts commit 37b7d05aed5dc90f6d5593325b6eb77494e21736. * update selector * tests * Rework number equality Compare concrete numbers, so we can get actual numbers in the error message when they are not equal, not some uint64 representation. This also uses InDelta for comparing floats. * Ensure that Observers are registered in the same order * Run observers in fixed order So the tests can be reproducible - iterating a map made the order of measurements random. * Ensure the proper alignment of the delegates This wasn't checked at all. After adding the checks, the test-386 failed. * Small tweaks to the global meter test * Ensure proper alignment of the callback pointer test-386 was complaining about it * update docs * update a TODO * address review issues * drop SetCallback Co-authored-by: Mauricio Vásquez <mauricio@kinvolk.io> Co-authored-by: Rahul Patel <rghetia@yahoo.com>
2020-03-05 12:15:30 -08:00
// lock protects `delegate` and `meters`.
lock sync.Mutex
// meters maintains a unique entry for every named Meter
// that has been registered through the global instance.
meters map[meterKey]*meterEntry
}
Add observer metric (#474) * wip: observers * wip: float observers * fix copy pasta * wip: rework observers in sdk * small fix in global meter * wip: aggregators and selectors * wip: monotonicity option for observers * some refactor * wip: docs needs more package docs (especially for api/metric and sdk/metric) * fix ci * Fix copy-pasta in docs Co-Authored-By: Mauricio Vásquez <mauricio@kinvolk.io> * recycle unused recorders in observers if a recorder for a labelset is unused for a second collection cycle in a row, drop it * unregister * thread-safe set callback * Fix docs * Revert "wip: aggregators and selectors" This reverts commit 37b7d05aed5dc90f6d5593325b6eb77494e21736. * update selector * tests * Rework number equality Compare concrete numbers, so we can get actual numbers in the error message when they are not equal, not some uint64 representation. This also uses InDelta for comparing floats. * Ensure that Observers are registered in the same order * Run observers in fixed order So the tests can be reproducible - iterating a map made the order of measurements random. * Ensure the proper alignment of the delegates This wasn't checked at all. After adding the checks, the test-386 failed. * Small tweaks to the global meter test * Ensure proper alignment of the callback pointer test-386 was complaining about it * update docs * update a TODO * address review issues * drop SetCallback Co-authored-by: Mauricio Vásquez <mauricio@kinvolk.io> Co-authored-by: Rahul Patel <rghetia@yahoo.com>
2020-03-05 12:15:30 -08:00
type meterImpl struct {
delegate unsafe.Pointer // (*metric.MeterImpl)
lock sync.Mutex
syncInsts []*syncImpl
asyncInsts []*asyncImpl
}
type meterEntry struct {
unique sdkapi.MeterImpl
impl meterImpl
}
type instrument struct {
descriptor sdkapi.Descriptor
Add observer metric (#474) * wip: observers * wip: float observers * fix copy pasta * wip: rework observers in sdk * small fix in global meter * wip: aggregators and selectors * wip: monotonicity option for observers * some refactor * wip: docs needs more package docs (especially for api/metric and sdk/metric) * fix ci * Fix copy-pasta in docs Co-Authored-By: Mauricio Vásquez <mauricio@kinvolk.io> * recycle unused recorders in observers if a recorder for a labelset is unused for a second collection cycle in a row, drop it * unregister * thread-safe set callback * Fix docs * Revert "wip: aggregators and selectors" This reverts commit 37b7d05aed5dc90f6d5593325b6eb77494e21736. * update selector * tests * Rework number equality Compare concrete numbers, so we can get actual numbers in the error message when they are not equal, not some uint64 representation. This also uses InDelta for comparing floats. * Ensure that Observers are registered in the same order * Run observers in fixed order So the tests can be reproducible - iterating a map made the order of measurements random. * Ensure the proper alignment of the delegates This wasn't checked at all. After adding the checks, the test-386 failed. * Small tweaks to the global meter test * Ensure proper alignment of the callback pointer test-386 was complaining about it * update docs * update a TODO * address review issues * drop SetCallback Co-authored-by: Mauricio Vásquez <mauricio@kinvolk.io> Co-authored-by: Rahul Patel <rghetia@yahoo.com>
2020-03-05 12:15:30 -08:00
}
type syncImpl struct {
delegate unsafe.Pointer // (*sdkapi.SyncImpl)
Add observer metric (#474) * wip: observers * wip: float observers * fix copy pasta * wip: rework observers in sdk * small fix in global meter * wip: aggregators and selectors * wip: monotonicity option for observers * some refactor * wip: docs needs more package docs (especially for api/metric and sdk/metric) * fix ci * Fix copy-pasta in docs Co-Authored-By: Mauricio Vásquez <mauricio@kinvolk.io> * recycle unused recorders in observers if a recorder for a labelset is unused for a second collection cycle in a row, drop it * unregister * thread-safe set callback * Fix docs * Revert "wip: aggregators and selectors" This reverts commit 37b7d05aed5dc90f6d5593325b6eb77494e21736. * update selector * tests * Rework number equality Compare concrete numbers, so we can get actual numbers in the error message when they are not equal, not some uint64 representation. This also uses InDelta for comparing floats. * Ensure that Observers are registered in the same order * Run observers in fixed order So the tests can be reproducible - iterating a map made the order of measurements random. * Ensure the proper alignment of the delegates This wasn't checked at all. After adding the checks, the test-386 failed. * Small tweaks to the global meter test * Ensure proper alignment of the callback pointer test-386 was complaining about it * update docs * update a TODO * address review issues * drop SetCallback Co-authored-by: Mauricio Vásquez <mauricio@kinvolk.io> Co-authored-by: Rahul Patel <rghetia@yahoo.com>
2020-03-05 12:15:30 -08:00
instrument
}
type asyncImpl struct {
delegate unsafe.Pointer // (*sdkapi.AsyncImpl)
instrument
runner sdkapi.AsyncRunner
Add observer metric (#474) * wip: observers * wip: float observers * fix copy pasta * wip: rework observers in sdk * small fix in global meter * wip: aggregators and selectors * wip: monotonicity option for observers * some refactor * wip: docs needs more package docs (especially for api/metric and sdk/metric) * fix ci * Fix copy-pasta in docs Co-Authored-By: Mauricio Vásquez <mauricio@kinvolk.io> * recycle unused recorders in observers if a recorder for a labelset is unused for a second collection cycle in a row, drop it * unregister * thread-safe set callback * Fix docs * Revert "wip: aggregators and selectors" This reverts commit 37b7d05aed5dc90f6d5593325b6eb77494e21736. * update selector * tests * Rework number equality Compare concrete numbers, so we can get actual numbers in the error message when they are not equal, not some uint64 representation. This also uses InDelta for comparing floats. * Ensure that Observers are registered in the same order * Run observers in fixed order So the tests can be reproducible - iterating a map made the order of measurements random. * Ensure the proper alignment of the delegates This wasn't checked at all. After adding the checks, the test-386 failed. * Small tweaks to the global meter test * Ensure proper alignment of the callback pointer test-386 was complaining about it * update docs * update a TODO * address review issues * drop SetCallback Co-authored-by: Mauricio Vásquez <mauricio@kinvolk.io> Co-authored-by: Rahul Patel <rghetia@yahoo.com>
2020-03-05 12:15:30 -08:00
}
var _ metric.MeterProvider = &meterProvider{}
var _ sdkapi.MeterImpl = &meterImpl{}
var _ sdkapi.InstrumentImpl = &syncImpl{}
var _ sdkapi.AsyncImpl = &asyncImpl{}
func (inst *instrument) Descriptor() sdkapi.Descriptor {
return inst.descriptor
}
// MeterProvider interface and delegation
func newMeterProvider() *meterProvider {
return &meterProvider{
meters: map[meterKey]*meterEntry{},
}
}
func (p *meterProvider) setDelegate(provider metric.MeterProvider) {
p.lock.Lock()
defer p.lock.Unlock()
p.delegate = provider
for key, entry := range p.meters {
Separate InstrumentationLibrary from metric.Descriptor (#2197) * factor instrumentation library out of the instrument descriptor * SDK tests pass * checkpoint work * otlp and opencensus tests passing * prometheus * tests pass, working on lint * lint applied: MetricReader->Reader * comments * Changelog * Apply suggestions from code review Co-authored-by: alrex <alrex.boten@gmail.com> * remove an interdependency * fix build * re-indent one * Apply suggestions from code review Co-authored-by: Tyler Yahn <MrAlias@users.noreply.github.com> * Lint&feedback * update after rename * comment fix * style fix for meter options * remove libraryReader, let Controller implement the reader API directly * rename 'impl' field to 'provider' * remove a type assertion * move metric/registry into internal; move registry.MeterProvider into metric controller * add test for controller registry function * CheckpointSet->Reader everywhere * lint * remove two unnecessary accessor methods; Controller implements MeterProvider and InstrumentationLibraryReader directly, no need to get these * use a sync.Map * ensure the initOnce is always called; handle multiple errors * Apply suggestions from code review Co-authored-by: Anthony Mirabella <a9@aneurysm9.com> * cleanup locking in metrictest * Revert "ensure the initOnce is always called; handle multiple errors" This reverts commit 3356eb5ed0c288ac3edcc2bc2e853aecda7f29b3. * Revert "use a sync.Map" This reverts commit ea7bc599bd3a24c4acb4cd9facd13f08cd702237. * restore the TODO about sync.Map Co-authored-by: alrex <alrex.boten@gmail.com> Co-authored-by: Tyler Yahn <MrAlias@users.noreply.github.com> Co-authored-by: Anthony Mirabella <a9@aneurysm9.com>
2021-09-27 08:51:47 -07:00
entry.impl.setDelegate(key, provider)
}
p.meters = nil
}
func (p *meterProvider) Meter(instrumentationName string, opts ...metric.MeterOption) metric.Meter {
p.lock.Lock()
defer p.lock.Unlock()
if p.delegate != nil {
return p.delegate.Meter(instrumentationName, opts...)
}
Separate InstrumentationLibrary from metric.Descriptor (#2197) * factor instrumentation library out of the instrument descriptor * SDK tests pass * checkpoint work * otlp and opencensus tests passing * prometheus * tests pass, working on lint * lint applied: MetricReader->Reader * comments * Changelog * Apply suggestions from code review Co-authored-by: alrex <alrex.boten@gmail.com> * remove an interdependency * fix build * re-indent one * Apply suggestions from code review Co-authored-by: Tyler Yahn <MrAlias@users.noreply.github.com> * Lint&feedback * update after rename * comment fix * style fix for meter options * remove libraryReader, let Controller implement the reader API directly * rename 'impl' field to 'provider' * remove a type assertion * move metric/registry into internal; move registry.MeterProvider into metric controller * add test for controller registry function * CheckpointSet->Reader everywhere * lint * remove two unnecessary accessor methods; Controller implements MeterProvider and InstrumentationLibraryReader directly, no need to get these * use a sync.Map * ensure the initOnce is always called; handle multiple errors * Apply suggestions from code review Co-authored-by: Anthony Mirabella <a9@aneurysm9.com> * cleanup locking in metrictest * Revert "ensure the initOnce is always called; handle multiple errors" This reverts commit 3356eb5ed0c288ac3edcc2bc2e853aecda7f29b3. * Revert "use a sync.Map" This reverts commit ea7bc599bd3a24c4acb4cd9facd13f08cd702237. * restore the TODO about sync.Map Co-authored-by: alrex <alrex.boten@gmail.com> Co-authored-by: Tyler Yahn <MrAlias@users.noreply.github.com> Co-authored-by: Anthony Mirabella <a9@aneurysm9.com>
2021-09-27 08:51:47 -07:00
cfg := metric.NewMeterConfig(opts...)
key := meterKey{
Separate InstrumentationLibrary from metric.Descriptor (#2197) * factor instrumentation library out of the instrument descriptor * SDK tests pass * checkpoint work * otlp and opencensus tests passing * prometheus * tests pass, working on lint * lint applied: MetricReader->Reader * comments * Changelog * Apply suggestions from code review Co-authored-by: alrex <alrex.boten@gmail.com> * remove an interdependency * fix build * re-indent one * Apply suggestions from code review Co-authored-by: Tyler Yahn <MrAlias@users.noreply.github.com> * Lint&feedback * update after rename * comment fix * style fix for meter options * remove libraryReader, let Controller implement the reader API directly * rename 'impl' field to 'provider' * remove a type assertion * move metric/registry into internal; move registry.MeterProvider into metric controller * add test for controller registry function * CheckpointSet->Reader everywhere * lint * remove two unnecessary accessor methods; Controller implements MeterProvider and InstrumentationLibraryReader directly, no need to get these * use a sync.Map * ensure the initOnce is always called; handle multiple errors * Apply suggestions from code review Co-authored-by: Anthony Mirabella <a9@aneurysm9.com> * cleanup locking in metrictest * Revert "ensure the initOnce is always called; handle multiple errors" This reverts commit 3356eb5ed0c288ac3edcc2bc2e853aecda7f29b3. * Revert "use a sync.Map" This reverts commit ea7bc599bd3a24c4acb4cd9facd13f08cd702237. * restore the TODO about sync.Map Co-authored-by: alrex <alrex.boten@gmail.com> Co-authored-by: Tyler Yahn <MrAlias@users.noreply.github.com> Co-authored-by: Anthony Mirabella <a9@aneurysm9.com>
2021-09-27 08:51:47 -07:00
InstrumentationName: instrumentationName,
InstrumentationVersion: cfg.InstrumentationVersion(),
SchemaURL: cfg.SchemaURL(),
}
entry, ok := p.meters[key]
if !ok {
entry = &meterEntry{}
Separate InstrumentationLibrary from metric.Descriptor (#2197) * factor instrumentation library out of the instrument descriptor * SDK tests pass * checkpoint work * otlp and opencensus tests passing * prometheus * tests pass, working on lint * lint applied: MetricReader->Reader * comments * Changelog * Apply suggestions from code review Co-authored-by: alrex <alrex.boten@gmail.com> * remove an interdependency * fix build * re-indent one * Apply suggestions from code review Co-authored-by: Tyler Yahn <MrAlias@users.noreply.github.com> * Lint&feedback * update after rename * comment fix * style fix for meter options * remove libraryReader, let Controller implement the reader API directly * rename 'impl' field to 'provider' * remove a type assertion * move metric/registry into internal; move registry.MeterProvider into metric controller * add test for controller registry function * CheckpointSet->Reader everywhere * lint * remove two unnecessary accessor methods; Controller implements MeterProvider and InstrumentationLibraryReader directly, no need to get these * use a sync.Map * ensure the initOnce is always called; handle multiple errors * Apply suggestions from code review Co-authored-by: Anthony Mirabella <a9@aneurysm9.com> * cleanup locking in metrictest * Revert "ensure the initOnce is always called; handle multiple errors" This reverts commit 3356eb5ed0c288ac3edcc2bc2e853aecda7f29b3. * Revert "use a sync.Map" This reverts commit ea7bc599bd3a24c4acb4cd9facd13f08cd702237. * restore the TODO about sync.Map Co-authored-by: alrex <alrex.boten@gmail.com> Co-authored-by: Tyler Yahn <MrAlias@users.noreply.github.com> Co-authored-by: Anthony Mirabella <a9@aneurysm9.com>
2021-09-27 08:51:47 -07:00
// Note: This code implements its own MeterProvider
// name-uniqueness logic because there is
// synchronization required at the moment of
// delegation. We use the same instrument-uniqueness
// checking the real SDK uses here:
entry.unique = registry.NewUniqueInstrumentMeterImpl(&entry.impl)
p.meters[key] = entry
}
Separate InstrumentationLibrary from metric.Descriptor (#2197) * factor instrumentation library out of the instrument descriptor * SDK tests pass * checkpoint work * otlp and opencensus tests passing * prometheus * tests pass, working on lint * lint applied: MetricReader->Reader * comments * Changelog * Apply suggestions from code review Co-authored-by: alrex <alrex.boten@gmail.com> * remove an interdependency * fix build * re-indent one * Apply suggestions from code review Co-authored-by: Tyler Yahn <MrAlias@users.noreply.github.com> * Lint&feedback * update after rename * comment fix * style fix for meter options * remove libraryReader, let Controller implement the reader API directly * rename 'impl' field to 'provider' * remove a type assertion * move metric/registry into internal; move registry.MeterProvider into metric controller * add test for controller registry function * CheckpointSet->Reader everywhere * lint * remove two unnecessary accessor methods; Controller implements MeterProvider and InstrumentationLibraryReader directly, no need to get these * use a sync.Map * ensure the initOnce is always called; handle multiple errors * Apply suggestions from code review Co-authored-by: Anthony Mirabella <a9@aneurysm9.com> * cleanup locking in metrictest * Revert "ensure the initOnce is always called; handle multiple errors" This reverts commit 3356eb5ed0c288ac3edcc2bc2e853aecda7f29b3. * Revert "use a sync.Map" This reverts commit ea7bc599bd3a24c4acb4cd9facd13f08cd702237. * restore the TODO about sync.Map Co-authored-by: alrex <alrex.boten@gmail.com> Co-authored-by: Tyler Yahn <MrAlias@users.noreply.github.com> Co-authored-by: Anthony Mirabella <a9@aneurysm9.com>
2021-09-27 08:51:47 -07:00
return metric.WrapMeterImpl(entry.unique)
}
// Meter interface and delegation
Separate InstrumentationLibrary from metric.Descriptor (#2197) * factor instrumentation library out of the instrument descriptor * SDK tests pass * checkpoint work * otlp and opencensus tests passing * prometheus * tests pass, working on lint * lint applied: MetricReader->Reader * comments * Changelog * Apply suggestions from code review Co-authored-by: alrex <alrex.boten@gmail.com> * remove an interdependency * fix build * re-indent one * Apply suggestions from code review Co-authored-by: Tyler Yahn <MrAlias@users.noreply.github.com> * Lint&feedback * update after rename * comment fix * style fix for meter options * remove libraryReader, let Controller implement the reader API directly * rename 'impl' field to 'provider' * remove a type assertion * move metric/registry into internal; move registry.MeterProvider into metric controller * add test for controller registry function * CheckpointSet->Reader everywhere * lint * remove two unnecessary accessor methods; Controller implements MeterProvider and InstrumentationLibraryReader directly, no need to get these * use a sync.Map * ensure the initOnce is always called; handle multiple errors * Apply suggestions from code review Co-authored-by: Anthony Mirabella <a9@aneurysm9.com> * cleanup locking in metrictest * Revert "ensure the initOnce is always called; handle multiple errors" This reverts commit 3356eb5ed0c288ac3edcc2bc2e853aecda7f29b3. * Revert "use a sync.Map" This reverts commit ea7bc599bd3a24c4acb4cd9facd13f08cd702237. * restore the TODO about sync.Map Co-authored-by: alrex <alrex.boten@gmail.com> Co-authored-by: Tyler Yahn <MrAlias@users.noreply.github.com> Co-authored-by: Anthony Mirabella <a9@aneurysm9.com>
2021-09-27 08:51:47 -07:00
func (m *meterImpl) setDelegate(key meterKey, provider metric.MeterProvider) {
m.lock.Lock()
defer m.lock.Unlock()
d := new(sdkapi.MeterImpl)
Separate InstrumentationLibrary from metric.Descriptor (#2197) * factor instrumentation library out of the instrument descriptor * SDK tests pass * checkpoint work * otlp and opencensus tests passing * prometheus * tests pass, working on lint * lint applied: MetricReader->Reader * comments * Changelog * Apply suggestions from code review Co-authored-by: alrex <alrex.boten@gmail.com> * remove an interdependency * fix build * re-indent one * Apply suggestions from code review Co-authored-by: Tyler Yahn <MrAlias@users.noreply.github.com> * Lint&feedback * update after rename * comment fix * style fix for meter options * remove libraryReader, let Controller implement the reader API directly * rename 'impl' field to 'provider' * remove a type assertion * move metric/registry into internal; move registry.MeterProvider into metric controller * add test for controller registry function * CheckpointSet->Reader everywhere * lint * remove two unnecessary accessor methods; Controller implements MeterProvider and InstrumentationLibraryReader directly, no need to get these * use a sync.Map * ensure the initOnce is always called; handle multiple errors * Apply suggestions from code review Co-authored-by: Anthony Mirabella <a9@aneurysm9.com> * cleanup locking in metrictest * Revert "ensure the initOnce is always called; handle multiple errors" This reverts commit 3356eb5ed0c288ac3edcc2bc2e853aecda7f29b3. * Revert "use a sync.Map" This reverts commit ea7bc599bd3a24c4acb4cd9facd13f08cd702237. * restore the TODO about sync.Map Co-authored-by: alrex <alrex.boten@gmail.com> Co-authored-by: Tyler Yahn <MrAlias@users.noreply.github.com> Co-authored-by: Anthony Mirabella <a9@aneurysm9.com>
2021-09-27 08:51:47 -07:00
*d = provider.Meter(
key.InstrumentationName,
metric.WithInstrumentationVersion(key.InstrumentationVersion),
metric.WithSchemaURL(key.SchemaURL),
).MeterImpl()
m.delegate = unsafe.Pointer(d)
for _, inst := range m.syncInsts {
inst.setDelegate(*d)
}
m.syncInsts = nil
for _, obs := range m.asyncInsts {
obs.setDelegate(*d)
Add observer metric (#474) * wip: observers * wip: float observers * fix copy pasta * wip: rework observers in sdk * small fix in global meter * wip: aggregators and selectors * wip: monotonicity option for observers * some refactor * wip: docs needs more package docs (especially for api/metric and sdk/metric) * fix ci * Fix copy-pasta in docs Co-Authored-By: Mauricio Vásquez <mauricio@kinvolk.io> * recycle unused recorders in observers if a recorder for a labelset is unused for a second collection cycle in a row, drop it * unregister * thread-safe set callback * Fix docs * Revert "wip: aggregators and selectors" This reverts commit 37b7d05aed5dc90f6d5593325b6eb77494e21736. * update selector * tests * Rework number equality Compare concrete numbers, so we can get actual numbers in the error message when they are not equal, not some uint64 representation. This also uses InDelta for comparing floats. * Ensure that Observers are registered in the same order * Run observers in fixed order So the tests can be reproducible - iterating a map made the order of measurements random. * Ensure the proper alignment of the delegates This wasn't checked at all. After adding the checks, the test-386 failed. * Small tweaks to the global meter test * Ensure proper alignment of the callback pointer test-386 was complaining about it * update docs * update a TODO * address review issues * drop SetCallback Co-authored-by: Mauricio Vásquez <mauricio@kinvolk.io> Co-authored-by: Rahul Patel <rghetia@yahoo.com>
2020-03-05 12:15:30 -08:00
}
m.asyncInsts = nil
}
func (m *meterImpl) NewSyncInstrument(desc sdkapi.Descriptor) (sdkapi.SyncImpl, error) {
m.lock.Lock()
defer m.lock.Unlock()
if meterPtr := (*sdkapi.MeterImpl)(atomic.LoadPointer(&m.delegate)); meterPtr != nil {
return (*meterPtr).NewSyncInstrument(desc)
}
inst := &syncImpl{
instrument: instrument{
descriptor: desc,
},
}
m.syncInsts = append(m.syncInsts, inst)
return inst, nil
}
// Synchronous delegation
func (inst *syncImpl) setDelegate(d sdkapi.MeterImpl) {
implPtr := new(sdkapi.SyncImpl)
var err error
*implPtr, err = d.NewSyncInstrument(inst.descriptor)
if err != nil {
// TODO: There is no standard way to deliver this error to the user.
// See https://github.com/open-telemetry/opentelemetry-go/issues/514
// Note that the default SDK will not generate any errors yet, this is
// only for added safety.
panic(err)
}
atomic.StorePointer(&inst.delegate, unsafe.Pointer(implPtr))
}
func (inst *syncImpl) Implementation() interface{} {
if implPtr := (*sdkapi.SyncImpl)(atomic.LoadPointer(&inst.delegate)); implPtr != nil {
return (*implPtr).Implementation()
}
return inst
}
// Async delegation
Add observer metric (#474) * wip: observers * wip: float observers * fix copy pasta * wip: rework observers in sdk * small fix in global meter * wip: aggregators and selectors * wip: monotonicity option for observers * some refactor * wip: docs needs more package docs (especially for api/metric and sdk/metric) * fix ci * Fix copy-pasta in docs Co-Authored-By: Mauricio Vásquez <mauricio@kinvolk.io> * recycle unused recorders in observers if a recorder for a labelset is unused for a second collection cycle in a row, drop it * unregister * thread-safe set callback * Fix docs * Revert "wip: aggregators and selectors" This reverts commit 37b7d05aed5dc90f6d5593325b6eb77494e21736. * update selector * tests * Rework number equality Compare concrete numbers, so we can get actual numbers in the error message when they are not equal, not some uint64 representation. This also uses InDelta for comparing floats. * Ensure that Observers are registered in the same order * Run observers in fixed order So the tests can be reproducible - iterating a map made the order of measurements random. * Ensure the proper alignment of the delegates This wasn't checked at all. After adding the checks, the test-386 failed. * Small tweaks to the global meter test * Ensure proper alignment of the callback pointer test-386 was complaining about it * update docs * update a TODO * address review issues * drop SetCallback Co-authored-by: Mauricio Vásquez <mauricio@kinvolk.io> Co-authored-by: Rahul Patel <rghetia@yahoo.com>
2020-03-05 12:15:30 -08:00
func (m *meterImpl) NewAsyncInstrument(
desc sdkapi.Descriptor,
runner sdkapi.AsyncRunner,
) (sdkapi.AsyncImpl, error) {
m.lock.Lock()
defer m.lock.Unlock()
Add observer metric (#474) * wip: observers * wip: float observers * fix copy pasta * wip: rework observers in sdk * small fix in global meter * wip: aggregators and selectors * wip: monotonicity option for observers * some refactor * wip: docs needs more package docs (especially for api/metric and sdk/metric) * fix ci * Fix copy-pasta in docs Co-Authored-By: Mauricio Vásquez <mauricio@kinvolk.io> * recycle unused recorders in observers if a recorder for a labelset is unused for a second collection cycle in a row, drop it * unregister * thread-safe set callback * Fix docs * Revert "wip: aggregators and selectors" This reverts commit 37b7d05aed5dc90f6d5593325b6eb77494e21736. * update selector * tests * Rework number equality Compare concrete numbers, so we can get actual numbers in the error message when they are not equal, not some uint64 representation. This also uses InDelta for comparing floats. * Ensure that Observers are registered in the same order * Run observers in fixed order So the tests can be reproducible - iterating a map made the order of measurements random. * Ensure the proper alignment of the delegates This wasn't checked at all. After adding the checks, the test-386 failed. * Small tweaks to the global meter test * Ensure proper alignment of the callback pointer test-386 was complaining about it * update docs * update a TODO * address review issues * drop SetCallback Co-authored-by: Mauricio Vásquez <mauricio@kinvolk.io> Co-authored-by: Rahul Patel <rghetia@yahoo.com>
2020-03-05 12:15:30 -08:00
if meterPtr := (*sdkapi.MeterImpl)(atomic.LoadPointer(&m.delegate)); meterPtr != nil {
return (*meterPtr).NewAsyncInstrument(desc, runner)
}
inst := &asyncImpl{
instrument: instrument{
descriptor: desc,
},
runner: runner,
Add observer metric (#474) * wip: observers * wip: float observers * fix copy pasta * wip: rework observers in sdk * small fix in global meter * wip: aggregators and selectors * wip: monotonicity option for observers * some refactor * wip: docs needs more package docs (especially for api/metric and sdk/metric) * fix ci * Fix copy-pasta in docs Co-Authored-By: Mauricio Vásquez <mauricio@kinvolk.io> * recycle unused recorders in observers if a recorder for a labelset is unused for a second collection cycle in a row, drop it * unregister * thread-safe set callback * Fix docs * Revert "wip: aggregators and selectors" This reverts commit 37b7d05aed5dc90f6d5593325b6eb77494e21736. * update selector * tests * Rework number equality Compare concrete numbers, so we can get actual numbers in the error message when they are not equal, not some uint64 representation. This also uses InDelta for comparing floats. * Ensure that Observers are registered in the same order * Run observers in fixed order So the tests can be reproducible - iterating a map made the order of measurements random. * Ensure the proper alignment of the delegates This wasn't checked at all. After adding the checks, the test-386 failed. * Small tweaks to the global meter test * Ensure proper alignment of the callback pointer test-386 was complaining about it * update docs * update a TODO * address review issues * drop SetCallback Co-authored-by: Mauricio Vásquez <mauricio@kinvolk.io> Co-authored-by: Rahul Patel <rghetia@yahoo.com>
2020-03-05 12:15:30 -08:00
}
m.asyncInsts = append(m.asyncInsts, inst)
return inst, nil
Add observer metric (#474) * wip: observers * wip: float observers * fix copy pasta * wip: rework observers in sdk * small fix in global meter * wip: aggregators and selectors * wip: monotonicity option for observers * some refactor * wip: docs needs more package docs (especially for api/metric and sdk/metric) * fix ci * Fix copy-pasta in docs Co-Authored-By: Mauricio Vásquez <mauricio@kinvolk.io> * recycle unused recorders in observers if a recorder for a labelset is unused for a second collection cycle in a row, drop it * unregister * thread-safe set callback * Fix docs * Revert "wip: aggregators and selectors" This reverts commit 37b7d05aed5dc90f6d5593325b6eb77494e21736. * update selector * tests * Rework number equality Compare concrete numbers, so we can get actual numbers in the error message when they are not equal, not some uint64 representation. This also uses InDelta for comparing floats. * Ensure that Observers are registered in the same order * Run observers in fixed order So the tests can be reproducible - iterating a map made the order of measurements random. * Ensure the proper alignment of the delegates This wasn't checked at all. After adding the checks, the test-386 failed. * Small tweaks to the global meter test * Ensure proper alignment of the callback pointer test-386 was complaining about it * update docs * update a TODO * address review issues * drop SetCallback Co-authored-by: Mauricio Vásquez <mauricio@kinvolk.io> Co-authored-by: Rahul Patel <rghetia@yahoo.com>
2020-03-05 12:15:30 -08:00
}
func (obs *asyncImpl) Implementation() interface{} {
if implPtr := (*sdkapi.AsyncImpl)(atomic.LoadPointer(&obs.delegate)); implPtr != nil {
return (*implPtr).Implementation()
}
return obs
Add observer metric (#474) * wip: observers * wip: float observers * fix copy pasta * wip: rework observers in sdk * small fix in global meter * wip: aggregators and selectors * wip: monotonicity option for observers * some refactor * wip: docs needs more package docs (especially for api/metric and sdk/metric) * fix ci * Fix copy-pasta in docs Co-Authored-By: Mauricio Vásquez <mauricio@kinvolk.io> * recycle unused recorders in observers if a recorder for a labelset is unused for a second collection cycle in a row, drop it * unregister * thread-safe set callback * Fix docs * Revert "wip: aggregators and selectors" This reverts commit 37b7d05aed5dc90f6d5593325b6eb77494e21736. * update selector * tests * Rework number equality Compare concrete numbers, so we can get actual numbers in the error message when they are not equal, not some uint64 representation. This also uses InDelta for comparing floats. * Ensure that Observers are registered in the same order * Run observers in fixed order So the tests can be reproducible - iterating a map made the order of measurements random. * Ensure the proper alignment of the delegates This wasn't checked at all. After adding the checks, the test-386 failed. * Small tweaks to the global meter test * Ensure proper alignment of the callback pointer test-386 was complaining about it * update docs * update a TODO * address review issues * drop SetCallback Co-authored-by: Mauricio Vásquez <mauricio@kinvolk.io> Co-authored-by: Rahul Patel <rghetia@yahoo.com>
2020-03-05 12:15:30 -08:00
}
func (obs *asyncImpl) setDelegate(d sdkapi.MeterImpl) {
implPtr := new(sdkapi.AsyncImpl)
var err error
*implPtr, err = d.NewAsyncInstrument(obs.descriptor, obs.runner)
if err != nil {
// TODO: There is no standard way to deliver this error to the user.
// See https://github.com/open-telemetry/opentelemetry-go/issues/514
// Note that the default SDK will not generate any errors yet, this is
// only for added safety.
panic(err)
}
atomic.StorePointer(&obs.delegate, unsafe.Pointer(implPtr))
Add observer metric (#474) * wip: observers * wip: float observers * fix copy pasta * wip: rework observers in sdk * small fix in global meter * wip: aggregators and selectors * wip: monotonicity option for observers * some refactor * wip: docs needs more package docs (especially for api/metric and sdk/metric) * fix ci * Fix copy-pasta in docs Co-Authored-By: Mauricio Vásquez <mauricio@kinvolk.io> * recycle unused recorders in observers if a recorder for a labelset is unused for a second collection cycle in a row, drop it * unregister * thread-safe set callback * Fix docs * Revert "wip: aggregators and selectors" This reverts commit 37b7d05aed5dc90f6d5593325b6eb77494e21736. * update selector * tests * Rework number equality Compare concrete numbers, so we can get actual numbers in the error message when they are not equal, not some uint64 representation. This also uses InDelta for comparing floats. * Ensure that Observers are registered in the same order * Run observers in fixed order So the tests can be reproducible - iterating a map made the order of measurements random. * Ensure the proper alignment of the delegates This wasn't checked at all. After adding the checks, the test-386 failed. * Small tweaks to the global meter test * Ensure proper alignment of the callback pointer test-386 was complaining about it * update docs * update a TODO * address review issues * drop SetCallback Co-authored-by: Mauricio Vásquez <mauricio@kinvolk.io> Co-authored-by: Rahul Patel <rghetia@yahoo.com>
2020-03-05 12:15:30 -08:00
}
// Metric updates
func (m *meterImpl) RecordBatch(ctx context.Context, labels []attribute.KeyValue, measurements ...sdkapi.Measurement) {
if delegatePtr := (*sdkapi.MeterImpl)(atomic.LoadPointer(&m.delegate)); delegatePtr != nil {
(*delegatePtr).RecordBatch(ctx, labels, measurements...)
}
}
func (inst *syncImpl) RecordOne(ctx context.Context, number number.Number, labels []attribute.KeyValue) {
if instPtr := (*sdkapi.SyncImpl)(atomic.LoadPointer(&inst.delegate)); instPtr != nil {
(*instPtr).RecordOne(ctx, number, labels)
}
}
Add observer metric (#474) * wip: observers * wip: float observers * fix copy pasta * wip: rework observers in sdk * small fix in global meter * wip: aggregators and selectors * wip: monotonicity option for observers * some refactor * wip: docs needs more package docs (especially for api/metric and sdk/metric) * fix ci * Fix copy-pasta in docs Co-Authored-By: Mauricio Vásquez <mauricio@kinvolk.io> * recycle unused recorders in observers if a recorder for a labelset is unused for a second collection cycle in a row, drop it * unregister * thread-safe set callback * Fix docs * Revert "wip: aggregators and selectors" This reverts commit 37b7d05aed5dc90f6d5593325b6eb77494e21736. * update selector * tests * Rework number equality Compare concrete numbers, so we can get actual numbers in the error message when they are not equal, not some uint64 representation. This also uses InDelta for comparing floats. * Ensure that Observers are registered in the same order * Run observers in fixed order So the tests can be reproducible - iterating a map made the order of measurements random. * Ensure the proper alignment of the delegates This wasn't checked at all. After adding the checks, the test-386 failed. * Small tweaks to the global meter test * Ensure proper alignment of the callback pointer test-386 was complaining about it * update docs * update a TODO * address review issues * drop SetCallback Co-authored-by: Mauricio Vásquez <mauricio@kinvolk.io> Co-authored-by: Rahul Patel <rghetia@yahoo.com>
2020-03-05 12:15:30 -08:00
func AtomicFieldOffsets() map[string]uintptr {
return map[string]uintptr{
"meterProvider.delegate": unsafe.Offsetof(meterProvider{}.delegate),
"meterImpl.delegate": unsafe.Offsetof(meterImpl{}.delegate),
"syncImpl.delegate": unsafe.Offsetof(syncImpl{}.delegate),
"asyncImpl.delegate": unsafe.Offsetof(asyncImpl{}.delegate),
Add observer metric (#474) * wip: observers * wip: float observers * fix copy pasta * wip: rework observers in sdk * small fix in global meter * wip: aggregators and selectors * wip: monotonicity option for observers * some refactor * wip: docs needs more package docs (especially for api/metric and sdk/metric) * fix ci * Fix copy-pasta in docs Co-Authored-By: Mauricio Vásquez <mauricio@kinvolk.io> * recycle unused recorders in observers if a recorder for a labelset is unused for a second collection cycle in a row, drop it * unregister * thread-safe set callback * Fix docs * Revert "wip: aggregators and selectors" This reverts commit 37b7d05aed5dc90f6d5593325b6eb77494e21736. * update selector * tests * Rework number equality Compare concrete numbers, so we can get actual numbers in the error message when they are not equal, not some uint64 representation. This also uses InDelta for comparing floats. * Ensure that Observers are registered in the same order * Run observers in fixed order So the tests can be reproducible - iterating a map made the order of measurements random. * Ensure the proper alignment of the delegates This wasn't checked at all. After adding the checks, the test-386 failed. * Small tweaks to the global meter test * Ensure proper alignment of the callback pointer test-386 was complaining about it * update docs * update a TODO * address review issues * drop SetCallback Co-authored-by: Mauricio Vásquez <mauricio@kinvolk.io> Co-authored-by: Rahul Patel <rghetia@yahoo.com>
2020-03-05 12:15:30 -08:00
}
}