[SDK] BatchLogRecordProcessor::ForceFlush hangs for 10 seconds #2583
Description
Platform : Ubuntu 23.10
Build: opentelemetry 1.11.0 with WITH_ASYNC_EXPORT_PREVIEW enabled
Deps: curl 7.85.0
Here are the lines of code used to initialize the logger provider:
opentelemetry::exporter::otlp::OtlpHttpLogRecordExporterOptions opts;
auto exporter =
opentelemetry::exporter::otlp::OtlpHttpLogRecordExporterFactory::Create(
opts);
opentelemetry::sdk::logs::BatchLogRecordProcessorOptions batchOptions;
auto processor =
opentelemetry::sdk::logs::BatchLogRecordProcessorFactory::Create(
std::move(exporter),
batchOptions);
auto resource = opentelemetry::sdk::resource::Resource::Create(attributes);
std::shared_ptr<opentelemetry::logs::LoggerProvider> provider =
opentelemetry::sdk::logs::LoggerProviderFactory::Create(
std::move(processor),
resource);
opentelemetry::logs::Provider::SetLoggerProvider(provider);
Before quitting my application I do a ForceFlush with a timeout equal to 1 second.
auto loggerProvider =
reinterpret_cast<opentelemetry::sdk::logs::LoggerProvider*>(
provider);
loggerProvider->ForceFlush(std::chrono::seconds(1));
std::shared_ptr<opentelemetry::logs::LoggerProvider> none;
opentelemetry::logs::Provider::SetLoggerProvider(none);
When the network connection is not good, it can happen that my application hangs for 10 seconds, which corresponds to the timeout of the curl request which I don't want to change.
I'd like to know why the timeout is not taken into account.