Follow spec advice on span limits, batch processors

[jack-berg]

Working on #7023 to implement spec PR open-telemetry/opentelemetry-specification#4331, I realized other parts of our implementation are not aligned with the spec:

• SpanLimitsBuilder / LogLimitsBuilder enforce that all the values are positive, when the spec (rightly) says that 0 is an acceptable value. LogLimitsBuilder has the same issues.
• BatchSpanProcessorBuilder / BatchLogRecordProcessorBuilder don't enforce any limits around max queue size. This means its possible to set the queue size to zero or a negative value, which is not allowed in the spec. Not sure what the underlying queue implementations do with a zero or negative value, but we add checks for this. Also, the max export size must be less than or equal to the max queue size.

[manojks1999]

@jack-berg can I work on this

[jack-berg]

Go for it!

[devatbosch]

Hi @jack-berg ,

Hope you are doing good.

Just wanted to know, if you are expecting to get any help/general improvement across any part of the repository, I can help you with that.

Please do let me know, in-case of any such issue even its a bit difficult, meanwhile I try to find a few...

[jack-berg]

Hi thanks @devatbosch - take a look at the help wanted section in the readme and see if you're up for anything discussed there!

[onurkybsi]

Hey @jack-berg, I created a PR regarding this issue, would you have a chance to take a look at it?

[breedx-splk]

Also, the max export size must be less than or equal to the max queue size.

Why is this a requirement? Wouldn't it be ok to fill up an export batch with multiple queue drains?

[jack-berg]

@breedx-splk good point. We'll see what the spec folks say: open-telemetry/opentelemetry-specification#4388