What are you trying to achieve?
The OTLP exporter spec states the following concerning retry logic:
Transient errors MUST be handled with a retry strategy. This retry strategy MUST implement an exponential back-off with jitter to avoid overwhelming the destination until the network is restored or the destination has recovered.
There should be more specificity around how exactly the retry strategy should be configured with gRPC. Here's the proto definition for the gRPC Retry Policies. Without more specificity, language implementations may end up choosing arbitrary values for these parameters.
Perhaps there should also be configuration parameters for adjusting the retry parameters. Maybe retry can be enabled / disabled altogether, or maybe the various parameters can be tuned.
Additional context.
Discussing this in an opentelemetry-java issue.
What are you trying to achieve?
The OTLP exporter spec states the following concerning retry logic:
There should be more specificity around how exactly the retry strategy should be configured with gRPC. Here's the proto definition for the gRPC Retry Policies. Without more specificity, language implementations may end up choosing arbitrary values for these parameters.
Perhaps there should also be configuration parameters for adjusting the retry parameters. Maybe retry can be enabled / disabled altogether, or maybe the various parameters can be tuned.
Additional context.
Discussing this in an opentelemetry-java issue.