You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Requirement - what kind of business use case are you trying to solve?
When configured to send traces to my collector via HTTP, I would like my tracer to have a default batch size that matches the documented value. Ideally this batch should be a large-ish value so that the sender and collector aren't thrashing under load.
Problem - what in Jaeger blocks you from solving the requirement?
The HTTP sender appears to always override its default batch size of 100. It appears this was introduced in #323, but I don't see any comments on the relevant part of that diff nor any discussion referencing that pull request since.
Proposal - what do you suggest to solve the problem or improve the existing situation?
Omit the transport.HTTPBatchSize(1) configuration from the HTTPOption array created here
Any open questions to address
Is 100 a sane default value? I have zero intuition for this, and perhaps the batch size of 1 was done for some good reason. Given the lack of discussion around it, I suspect the author made that change for local testing, and it was merged in by accident.
The text was updated successfully, but these errors were encountered:
I am not sure what was the motivation for size=1 in that PR, maybe worth asking the author.
The typical min size of a span is around 500 bytes with common opentracing instrumentation, likely much larger with OpenTelemetry conventions. There's no particular wisdom in the default of 100.
This resolves#570 where HTTP spans were sent in batches of 1
instead of the documented default of 100.
Signed-off-by: R. Aidan Campbell <aidan.campbell@aexp.com>
Requirement - what kind of business use case are you trying to solve?
When configured to send traces to my collector via HTTP, I would like my tracer to have a default batch size that matches the documented value. Ideally this batch should be a large-ish value so that the sender and collector aren't thrashing under load.
Problem - what in Jaeger blocks you from solving the requirement?
The HTTP sender appears to always override its default batch size of 100. It appears this was introduced in #323, but I don't see any comments on the relevant part of that diff nor any discussion referencing that pull request since.
Proposal - what do you suggest to solve the problem or improve the existing situation?
Omit the
transport.HTTPBatchSize(1)
configuration from theHTTPOption
array created hereAny open questions to address
Is 100 a sane default value? I have zero intuition for this, and perhaps the batch size of 1 was done for some good reason. Given the lack of discussion around it, I suspect the author made that change for local testing, and it was merged in by accident.
The text was updated successfully, but these errors were encountered: