You've already forked opentelemetry-go
mirror of
https://github.com/open-telemetry/opentelemetry-go.git
synced 2025-11-25 22:41:46 +02:00
Clarify DefaultMaxQueueSize and DefaultScheduleDelay usage (#6974)
### Description
OpenTelemetry uses `DefaultScheduleDelay` and `DefaultExportTimeout`
values as milliseconds but Go time package will understand them as
nanoseconds.
I understand that this is a stable library and that those value will
probably never change, so can we at least clarify their usage?
Right above the defaults declaration it says `// Defaults for
BatchSpanProcessorOptions.` which is confusing.
We used `trace.DefaultScheduleDelay` as a fallback value for our tracing
setup.
This confusion led to high CPU usage due to the frequent batch exports.
### Confusing behavior
```go
processor := trace.NewBatchSpanProcessor(exporter,
// set timeout to 5000 ns instead of the expected 5000 ms
trace.WithBatchTimeout(trace.DefaultScheduleDelay),
// set timeout to 30000 ns instead of the expected 30000 ms
trace.WithExportTimeout(trace.DefaultExportTimeout),
)
```
### Correct way to use those values
```go
processor := trace.NewBatchSpanProcessor(exporter,
trace.WithBatchTimeout(trace.DefaultScheduleDelay * time.Millisecond),
trace.WithExportTimeout(trace.DefaultExportTimeout * time.Millisecond),
)
```
---------
Co-authored-by: Damien Mathieu <42@dmathieu.com>
Co-authored-by: Robert Pająk <pellared@hotmail.com>
This commit is contained in:
@@ -18,8 +18,10 @@ import (
|
||||
|
||||
// Defaults for BatchSpanProcessorOptions.
|
||||
const (
|
||||
DefaultMaxQueueSize = 2048
|
||||
DefaultScheduleDelay = 5000
|
||||
DefaultMaxQueueSize = 2048
|
||||
// DefaultScheduleDelay is the delay interval between two consecutive exports, in milliseconds.
|
||||
DefaultScheduleDelay = 5000
|
||||
// DefaultExportTimeout is the duration after which an export is cancelled, in milliseconds.
|
||||
DefaultExportTimeout = 30000
|
||||
DefaultMaxExportBatchSize = 512
|
||||
)
|
||||
|
||||
Reference in New Issue
Block a user